Proceedings of the 5th International Conference on Applied Human Factors and Ergonomics AHFE 2014, Kraków, Poland 19-23 July 2014 Edited by T. Ahram, W. Karwowski and T. Marek How to Design Great Emotional Experience for Mobile User Interface? Dongbin Tobin Cho a , Jongwoo Jung a , Insik Myung a , Hyewon Lee a , and Patrick W. Jordan b a SAMSUNG ELECTRONICS Co., Ltd. 11, Seocho-daero 74-gil Seocho-gu, Seoul 137-965, Korea b Royal Society of Medicine Candlesby House Butleigh, Somerset BA6 8TH, UK ABSTRACT Emotion is easily recognized but difficult to define. For example, emotion has been a famous topic for a long time in user experience (UX) design. However, it has been a question of long standing how to systematically evaluate and design emotion, while maintaining qualitative aspects of emotion (e.g., not-measurable aspects). We developed the systematic tool for emotional UX design, while leaving some room for designer’s creativity. To develop the tool, participants performed various tasks using smartphone, generating both positive and negative emotions. Once people’s emotions were identified, ideal emotional journey, consisting emotional solutions, were created. In order to create ideal emotions, general design principles were proposed. Finally, more specific UX solutions were developed based on general design principles. Some of our UX solutions were applied to design new messenger for smartphone, emotionally augmenting communicating experiences. Keywords: UX, Emotion, Design, Mobile, Smartphone, Solution INTRODUCTION What is Emotion? Emotions are a combination of mental and physical states. The mental side of emotion involves a psychological judgment about an experience, while the physical side comes from physiological reactions in the body caused in part by these psychological judgments and in part because of subconscious physiological reactions (Carlson and Hatfield, 1992). Emotions are not the only states that involve affect and activation. Desmet (2002) identified four other types. The states are differentiated from each other based on the length of time for which the affect and activation are experienced, on the complexity of the causes and whether they are directed at a specific situation or felt more generally. Emotions are compared and contrasted with the other four states, which are moods, sentiments, and personal traits (see Figure 1). Emotions tend to have a simple or single cause, be directed at something or someone specific and be comparatively short term, for example lasting seconds or minutes. For example, being annoyed with a friend because they were late for a meeting would be an example of an emotion. 5377
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Proceedings of the 5th International Conference on Applied Human Factors and Ergonomics AHFE 2014, Kraków, Poland 19-23 July 2014 Edited by T. Ahram, W. Karwowski and T. Marek
How to Design Great Emotional Experience for Mobile User Interface?
Dongbin Tobin Choa, Jongwoo Junga, Insik Myunga, Hyewon Leea, and Patrick W. Jordan b
a SAMSUNG ELECTRONICS Co., Ltd.
11, Seocho-daero 74-gil Seocho-gu, Seoul 137-965, Korea
b Royal Society of Medicine
Candlesby House Butleigh, Somerset BA6 8TH, UK
ABSTRACT
Emotion is easily recognized but difficult to define. For example, emotion has been a famous topic for a long time in user experience (UX) design. However, it has been a question of long standing how to systematically evaluate and design emotion, while maintaining qualitative aspects of emotion (e.g., not-measurable aspects). We developed the systematic tool for emotional UX design, while leaving some room for designer’s creativity. To develop the tool, participants performed various tasks using smartphone, generating both positive and negative emotions. Once people’s emotions were identified, ideal emotional journey, consisting emotional solutions, were created. In order to create ideal emotions, general design principles were proposed. Finally, more specific UX solutions were developed based on general design principles. Some of our UX solutions were applied to design new messenger for smartphone, emotionally augmenting communicating experiences. Keywords: UX, Emotion, Design, Mobile, Smartphone, Solution
INTRODUCTION What is Emotion? Emotions are a combination of mental and physical states. The mental side of emotion involves a psychological judgment about an experience, while the physical side comes from physiological reactions in the body caused in part by these psychological judgments and in part because of subconscious physiological reactions (Carlson and Hatfield, 1992). Emotions are not the only states that involve affect and activation. Desmet (2002) identified four other types. The states are differentiated from each other based on the length of time for which the affect and activation are experienced, on the complexity of the causes and whether they are directed at a specific situation or felt more generally. Emotions are compared and contrasted with the other four states, which are moods, sentiments, and personal traits (see Figure 1). Emotions tend to have a simple or single cause, be directed at something or someone specific and be comparatively short term, for example lasting seconds or minutes. For example, being annoyed with a friend because they were late for a meeting would be an example of an emotion.
5377
Proceedings of the 5th International Conference on Applied Human Factors and Ergonomics AHFE 2014, Kraków, Poland 19-23 July 2014 Edited by T. Ahram, W. Karwowski and T. Marek
Moods are longer lasting than emotions, typically lasting for hours and possibly days. They often have combined causes and are felt generally rather than being directed only at the things that caused them. For instance, if our alarm clock does not go off we are late getting up, then our partner is rude to us and then we get stuck in traffic we may be in a bad mood by the time we arrive at work. Sentiments are affects and activations aimed at particular things over the long term. They reflect our likes, dislikes, attitudes and standards. While they are likely to influence our emotions, they are not the same thing. For example, someone might have the sentiment of being scared of snakes but will not actually feel the emotion of fear unless they see a snake (Frijda, 1994). When we are consistently predisposed to particular moods that we tend to return to over time this can be seen as an emotional personality trait. For example, if someone tends to be in a cheerful mood a lot we might characterize them as a cheerful person, while if someone is often in an angry mood we might describe them as an angry person.
Figure 1. What emotions are and are not This paper mainly involves the study on the emotional experiences (less on the moods, sentiments, or emotional personality traits), which tend to have simple or single cause of using smartphones. For instance, user might be emotionally satisfied because of clear feedback from the interface. Dimensions of Emotions Two basic dimensions of emotions are affect and activation. Nearly all models of emotion include these dimensions (e.g., Bales, 2001; Desmet, 2002; Mehrabian, 1981; Russell, 1980; Van Gorp, 2006). Affect, or positivity, refers to whether the emotion is positive or negative. Examples of emotions with positive affect are: excitement, happiness and being relaxed; examples of emotions with negative affect are irritation, disappointment and boredom. Some emotions are neutral in affect (neither positive nor negative). Examples of these include caution and focus. Activation refers to how much energy or stimulation the emotion causes. For example, excitement and focus have quite high levels of activation, caution has a neutral level and relaxation and boredom low levels. One extra dimension that has been proposed is dominance (Mehrabian, 1980). This three dimensional model includes this in addition to affect and activation – they refer to these as ‘pleasure’ and ‘arousal’ and call their model the Pleasure, Arousal, Dominance (PAD) Model. For example, fear and anger are both emotions that are negative in affect and high in activation, but they are different in terms of dominance. Fear is an emotion resulting from a low dominance and a sense of not being in control or having any power, whereas anger is experienced when the person feels more dominant and powerful. (see Figure 4 for affect and activation, and Figure 5 for activation and dominance). These three dimensions were used to select final lists of 30 postive and 24 negative distinct emotions among many other similar emotions (see also RESULTS for developing emotion sets). Our Design Approach to Apply Emotion In order to optimize people’s user experience with products it is important to ensure that they are not only useful and usable but also that they are emotionally engaging. Our emotions influence how we plan to use a product, how we interact with it and our perceptions surrounding the product before, during and after use (Forlizzi and Battarbee,
5378
Proceedings of the 5th International Conference on Applied Human Factors and Ergonomics AHFE 2014, Kraków, Poland 19-23 July 2014 Edited by T. Ahram, W. Karwowski and T. Marek
2004). Aside from strategic factors such as pricing, advertising and retail channels, emotional reactions to a product are probably the single biggest factor affecting its success in the marketplace (Van Gorp and Adams, 2012). Likewise, emotion has been important and well-known topic for a long time in user experience (UX) design. However, it has been a question how to systematically design the emotion, while maintaining qualitative aspects of emotion such as designer’s creativity and flexibility. Therefore, the main aim of the project was to create both systematic and flexible tools for emotional UX design and to show how this tool can be applied to mobile interface such as gallery and messenger. To achieve our goal, we followed several steps; 1. Identified emotions associated with smartphone use 2. Converted current emotions to ideal emotions 3. Delivered positive emotional experience using general solutions. 4. Delivered general solution using more specific UX design solution 5. Applied UX solutions to mobile interface with designer’s creativity and flexibility (see Table 1). Table 1: Steps to develop emotional interface
Steps Description Current Emotions (Emotional Journey)
User’s current emotional experience of using smartphones
Emotional Solutions Ideal emotions that boost up/solves the current emotional status/problems General Solutions High level design principle that, if followed, deliver the ideal emotions UX Solutions More specific design solution that, if followed, delver the general solutions. Emotional Design With selected UX solutions, designer can use his/her experiences and
creativity to design emotional interface.
METHOD Three empirical studies (user diaries, user trials, focus groups), plus a workshop and a subsequent series of work-sessions were carried out (immersion stage) (see Figure 2).
Figure 2. Overall research procedures
5379
Proceedings of the 5th International Conference on Applied Human Factors and Ergonomics AHFE 2014, Kraków, Poland 19-23 July 2014 Edited by T. Ahram, W. Karwowski and T. Marek
32 participants took part in the studies: 16 users of Android smartphones, 8 users iOS smartphones (Apple iPhone), and 8 users of other smartphones (Blackberry and Nokia). The participants were evenly balanced for age and gender and were also evenly distributed according to age and socioeconomic status. All participants took part in both the diary studies and the user trials, with 18—6 of each type of OS users—taking part in the focus groups. User Diary Study In user diary study, participants were asked to complete a series of smartphone tasks over the course of a week. They were asked to complete each task twice, once in the first half of the week and once in the second half. The task set included 13 tasks, such as make a telephone call, add someone’s contact details to the phone, and send a text by SMS or Instant Messenger, and etc. (see Table 2, for complete task lists). They completed the task ‘naturally’ (e.g., at home) as part of their smartphone usage, they were simply asked to record their experience of this (see Table 3, for complete questions) Table 2. Task set during diary study & user trials
Table 3. Items that participants were asked during diary study
5380
Proceedings of the 5th International Conference on Applied Human Factors and Ergonomics AHFE 2014, Kraków, Poland 19-23 July 2014 Edited by T. Ahram, W. Karwowski and T. Marek
User Trials In user trials, participants attended a usability lab in order to use a smartphone—the Samsung Galaxy S3—under controlled condition. They were given a selection of tasks from the same 13 that were done in the diary study (see Table 2). The order in which they were asked to do these was randomized and the number of tasks they did was dependent on how quickly they managed to complete them. The session lasted for 90 minutes. Users completed each task that they were set in silence, but were videoed as they did it. The video was then played back to them and they were asked to comment on how they were feeling at each stage of task, which technique is known as retrospective protocol. It has the advantage of enabling participants to complete the tasks without the intrusion of having to verbalize their emotions at the time and removes the risk that the effort of having to do this would make the tasks harder. Experimenter recorded the reactions such as participants’ positive and negative emotions as they went through the tasks (see Figure 3, for example). At the end of the trials participants were asked to provide more opinions about the best and worst aspects of the smartphone and the tasks that they found most and least enjoyable (see Table 4). Table 4. Information recorded during the user trials
Figure 3. Example of user’s emotional journey (task: add someone’s contact details to the phone) Focus Groups The focus groups were held involving 18 of the 32 people who had already participated in the diary and user trials studies. Each group contained six people who all used the same category of operating system (i.e., Android, iOS or ‘other’). Each group started with participants discussing the emotions associated with their favorite products. This was to give us an insight into whether there might be additional positive emotions that can be experienced in product use outside of those associated with smartphones. In the analysis this could be used as a check on the generalizability of the study outcomes to other product areas. They then discussed the emotions they felt when using their own smartphones and the Samsung Galaxy S3 that they had used in the user trials. They also discussed characteristics of the phones that were associated with these emotions. After the discussions, participants were
5381
Proceedings of the 5th International Conference on Applied Human Factors and Ergonomics AHFE 2014, Kraków, Poland 19-23 July 2014 Edited by T. Ahram, W. Karwowski and T. Marek
asked to record their own views with respect to each of the issues discussed (see Table 5) Table 5. Information recorded during the focus groups
Immersion In immersion stage, after completion of the empirical studies, some of the expert used the Samsung Galaxy S3 for a period of three weeks and reviewed the information gained from diary study, user trials, and focus groups. The aim was to give the project team real-life experience of using the smartphones which would generate insights to inform the analysis of the current emotions, the emotional solutions, general solutions, and UX solutions. An initial workshop, including the whole of the core project team, reviewed the information gained from the diaries, user trials, and focus groups and, on the basis of this, made an initial pass at identifying emotional and general solutions. In a series of subsequent work-sessions, the experts developed these further, to move towards the final set of emotional, general, UX solutions, and the relationships between UX and general solutions and between general and emotional solutions.
RESULTS Emotion Sets Based on the empirical tests and experts’ insight, 30 emotional solutions, 14 general solutions, and 23 UX solutions were developed. The emotion sets were produced in three stages. First, we listed all the separate emotional descriptors in users’ responses across the user diaries, user trials, and focus groups. This meant including all the emotions described at any stage during a task or when reflecting on a product. Second, we judged that some of the items that participants listed were not actually emotions. For example if, when asked about their emotions, someone had put ‘phone looks good’ or ‘difficult to use’ we would remove these items as they do not describe how the person is feeling, but rather are their assessments of the phone. After the first two stages, we found 67 of negative emotions and 87 of positive emotions. Third, experts grouped items together in instances where, in the context of product use, we judged them to be essentially the same. For example, although amazement and admiration have qualitative similarities—they are both about being impressed with the product—amazement is significantly greater than admiration in terms of positivity and activation and is lower on dominance (see Figure 5). Similarly, although anger and annoyance are both about the perception that some aspect of a product is unacceptably poor, anger is significantly greater in terms of negativity, activation and dominance. This led to our final lists of 24 distinct negative emotions and 30 distinct positive emotions (see Figure 4).
5382
Proceedings of the 5th International Conference on Applied Human Factors and Ergonomics AHFE 2014, Kraków, Poland 19-23 July 2014 Edited by T. Ahram, W. Karwowski and T. Marek
Figure 4. Emotions on activation and affect (positivity) dimensions Emotional solutions are particular positive emotions that we aim to elicit in order to help improve user experience during an emotional journey with a product. And the set of potential emotional solutions for smartphones is the same as the set of positive emotions that we identified during the studies (see Figure 5).
Figure 5. Emotional solutions (positive emotions) on activation and dominance dimensions General Solutions General solutions are high-level design principles for delivering the emotional solutions. They were based on linking the positive and negative emotions that had occurred in the studies to qualities and behaviors of the product and then generalizing from these to form the general solutions. In total there are 14 general solutions. Between them they can deliver all of the proposed emotional solutions as well as eliminating all of the potential negative emotions. They are listed in Table 6 with their subcomponents. To understand better with emotions, they are also plotted in the same dimension as emotions. For example, general solution of FLOW can give people feeling focused (compare
5383
Proceedings of the 5th International Conference on Applied Human Factors and Ergonomics AHFE 2014, Kraków, Poland 19-23 July 2014 Edited by T. Ahram, W. Karwowski and T. Marek
Figure 6 with Figure 5). Table 6. Lists of general solutions and their subcomponents
Figure 6. General solutions plotted on emotional dimensions (compare with Figure 5; e.g., FLOW gives feeling of focus) One of our general solutions is SUPERIOR QUALITY, which has several subcomponents. Since it is a general principle, it applies to both IT and non IT product, such as Lexus (see Figure 7). And the general solutions will form the basis for developing more specific UX solutions.
5384
Proceedings of the 5th International Conference on Applied Human Factors and Ergonomics AHFE 2014, Kraków, Poland 19-23 July 2014 Edited by T. Ahram, W. Karwowski and T. Marek
Figure 7. An example of general solution, SUPERIOR QUALITY UX Solution UX solutions represent design directions that embody principles behind a general solution. Based on general solutions, we developed 23 UX solutions which categorizes into ten groups (see Table 7). In sum, based on ideal emotions, general solutions were developed. And based on general solutions, UX solutions were developed. As shown in Figure 9 it describes the links between the UX solutions, general solutions, and the emotions experienced, showing systematically how UX solutions can enable the positive emotions and eliminate the negative.
5385
Proceedings of the 5th International Conference on Applied Human Factors and Ergonomics AHFE 2014, Kraków, Poland 19-23 July 2014 Edited by T. Ahram, W. Karwowski and T. Marek
Table 7. Lists of UX solutions and their general solution link
For example, one of our UX solution group is Perfecting Every Details, which has UX solutions of Natural Phenomena, Hidden Secrets and Beautifying Errors. These UX Solutions were developed from general solutions of SUPERIOR QUALITY, DEPENDABILITY, and CERTAINTY (see Figure 8).
Figure 8. Example of UX solutions which are based on general solution of SUPERIOR QUALITY (compare with Figure 7)
5386
Proceedings of the 5th International Conference on Applied Human Factors and Ergonomics AHFE 2014, Kraków, Poland 19-23 July 2014 Edited by T. Ahram, W. Karwowski and T. Marek
Messenger Design Example It is possible to apply emotional solutions, general solutions, and UX solutions to communication apps, such as messenger. As shown in Figure 9, for example, to develop “emotional messenger” for smartphone, we need to identify the emotions that user experience when using existing messenger (e.g., embarrassment, alienation, boredom, skepticism, nostalgia, indifference, etc., see Table 8). Table 8. Possible reasons for current emotions toward existing messenger
When current emotions are identified, ideal emotions are selected, which are already linked with negative or positive emotions (e.g. embarrassmentpride, skepticismanticipation). Then certain general solutions are selected to deliver the emotional solutions (e.g., general solution of SUPERIOR QUALITY gives feeling of pride; general solution of GRATIFYING NOVELTY gives feeling of anticipation). Finally, sets of UX solutions can be selected to deliver the general solution (e.g., UX solution of Natural Phenomena can generate the feeling of SUPERIOR QUALITY; UX solution of Unpredictability can generate GRATIFYING NOVELTY). When UX solutions of Natural Phenomena and Unpredictability are selected for new messenger design, designers can use their own creativity and experiences based on the suggested UX solutions. For instance, by using UX solution of Natural Phenomena, the space where communication is occurring can be transformed into real paper (e.g., real paper has natural quality of ink spreading, folding, air buffer, resilience, etc.). The resilient nature of paper sometimes creates pleasant Unpredictability for users such as applying pop-up-card effect to the messenger.
5387
Proceedings of the 5th International Conference on Applied Human Factors and Ergonomics AHFE 2014, Kraków, Poland 19-23 July 2014 Edited by T. Ahram, W. Karwowski and T. Marek
Figure 9. Possible steps to develop “emotional messenger”
CONCLUSION The main aim of this paper was to deliver a set of design process and principles for delivering a positive emotional experience with product and services. It also describes the systematic links between solutions and the emotions experienced, while maximizing designer’s creativity, flexibility, and previous experiences. Some of our concept design and prototypes were submitted as Grade-A patents, receiving certification of originality and creativity. Moreover, one of patents was directly applied to “Pen Window,” Samsung Galaxy Note 3. All of our outcomes were transferred to Mobile Communication Division, Samsung Electronics, which implies that our outcomes have great impact on mobile product.
REFERENCES Bales, R. F. (2001). “Social Interaction Systems: Theory and Measurement”, Transaction Publishers. Carlson, J. G., & Hatifield, E. (1992). “Psychology of Emotion”, Harcourt, Brace, Jovanovich. Desmet, P. R. (2002). “Designing Emotions”, Delft University of Technology, Department of Industrial Design. Forlizzi, J. & Battarbee, K. Understanding experience in interactive systems. Human-Computer Interaction Institute. Paper 46.
http://repository.cmu.edu/hcii/46. Frijda, N. H. (1994). Varieties of affect: Emotions and episodes, moods and sentiments. In Ekam, P. and Davidson, R. J.(eds),
“The Nature of Emotion, Fundamental Questions”, Oxford University Press. Mehrabian, A. (1980). “Basic Dimensions for a General Psychological Theory of Emotion”, Wadsworth Publishing Company. Mehrabian, A. (1981). “Silent Messages: Implicit Communication of Emotions and Attitudes”, Wadsworth Publishing Company. Russell, J.A. (1980). A circumplex model of affect. Journal of Personality and Social Psychology, 39, 1168-1178. Van Gorp, T. J. (2006). “Emotion, Arousal, Attention and Flow: Chaining Emotional States to Improve Human-Computer
Interaction”, University of Calgary Master’s Degree Project. Van Gorp, T. & Adams, E. (2012). “Design for Emotion”, Morgan Kaufmann; Elsevier Science.
5388
ORIGINAL ARTICLE
Object-based correspondence effects for action-relevantand surface-property judgments with keypress responses:evidence for a basis in spatial coding
Dongbin Tobin Cho • Robert W. Proctor
Received: 7 June 2012 / Accepted: 23 September 2012 / Published online: 26 October 2012
� Springer-Verlag Berlin Heidelberg 2012
Abstract It has been proposed that grasping affordances
produce a Simon-type correspondence effect for left–right
keypress responses and the location of the graspable part of
an object for judgments based on action-relevant properties
such as shape, but not on surface properties. We tested the
implications of this grasping affordance account and con-
trasted them with the ones derived from a spatial coding
account that distinguishes holistic processing of integral
dimensions and analytic processing of separable dimen-
sions. In Experiments 1–3, judgments about the color of a
door handle showed a Simon effect relative to the handle’s
base, whereas judgments about the handle’s shape showed
no Simon effect. In Experiment 4, when the middle of the
handle was colored, the Simon effect was obtained relative
to the base, but when the color was at the tip of the handle
or near the base, Simon effects were obtained relative to
the color location. For Experiment 5, only the base was
colored, and the Simon effect was larger for a passive
rather than active handle state, as in the color-judgment
conditions of Experiments 2–4 in which the colored region
overlapped with the base. In Experiment 6, orientation
judgments showed no Simon effect, as the shape judgments
did in Experiments 1 and 2. The findings of (a) an absence
of Simon effects for shape and orientation judgments,
(b) no larger Simon effects for active than passive handle
states, and (c) isolation of the changing component for
color judgments are consistent with the spatial coding
account, according to which the distinction between object
shape/orientation and color is one of integral versus sepa-
rable dimensions.
Introduction
The information-processing approach to human perfor-
mance assumes that perception and action are mediated
by representations, or codes. Therefore, specific behav-
ioral effects are attributed to the codes and processes that
underlie performance (Sanders, 1998). Recently, from
within the information-processing perspective, several
researchers have adopted the concept of affordance,
which originated in the ecological psychology approach
that espouses the view that perception is direct and not
mediated by representations (Gibson, 1979). Representa-
tional affordance accounts have been developed to
explain compatibility effects obtained with mappings of
objects with graspable properties to keypress or handgrip
Copyright of Psychological Research is the property of Springer Science & Business MediaB.V. and its content may not be copied or emailed to multiple sites or posted to a listservwithout the copyright holder's express written permission. However, users may print,download, or email articles for individual use.
Correspondence Effects for Objects With OpposingLeft and Right Protrusions
Dongbin (Tobin) Cho and Robert W. ProctorPurdue University
Choice reactions to a property of an object stimulus are often faster when the location of a graspable partof the object corresponds with the location of a keypress response than when it does not, a phenomenoncalled the object-based Simon effect. Experiments 1–3 examined this effect for variants of teapot stimulithat were oriented to the left or right. Whether keypress responses were made with fingers within thesame hand or between different hands was also manipulated. Experiment 1 showed that, for judgmentsof stimulus color and upright-inverted orientation, the Simon effect for intact teapots occurred in thedirection of the spout location and was larger for within- than between-hand response modes. InExperiments 2 and 3, teapots with the handle or spout removed showed separate contributions of eachcomponent to the Simon effect. In Experiment 4, we clarified a discrepancy between our findings ofobject-based Simon effects and a previously reported absence of effect with color judgments fordoor-handle stimuli. We obtained an object-based Simon effect with respect to handle position when thebases of the door handles were centered but not when the handles were centered. The findings thatobject-based Simon effects occur with color judgments and when responses are fingers on the same handare in closer agreement with a location coding account than with a grasping affordance account.
Keywords: affordance, grasping, Simon task, stimulus-response compatibility, within-hand responses
In stimulus-response compatibility (SRC) tasks, participants arerequired to make speeded response choices based on the values ofa stimulus property, often stimulus location. Performance of SRCtasks shows shorter reaction time (RT) for certain stimulus-response (S-R) mappings, called compatible, than for others, calledincompatible (Proctor & Vu, 2006). For spatial SRC tasks, re-sponses are faster and more accurate when the stimulus and responselocations correspond (e.g., left stimulus mapped to left response andright stimulus to right response) than when they do not.
Relative Location Coding
Because compatibility effects vary as a function of the concep-tual nature of the stimulus and response properties, they are, ingeneral, assumed to be a function of cognitive codes of thestimulus and response features (Proctor & Reeve, 1990). Forexample, regardless of whether responses are made to the right orto the left of the body midline, the spatial SRC effects are functionsof the relative positions of both the stimuli and the responses(Nicoletti, Anzola, Luppino, Rizzolatti, & Umilta, 1982). More-
over, the corresponding mappings of stimulus locations to re-sponse locations produce faster responses than do non-corresponding mappings, even when the hands are crossed so thatthe right hand makes the left response and the left hand the rightresponse (Anzola, Bertoloni, Buchtel, & Rizzolatti, 1977). Thisresult indicates that the compatibility effects are due to spatialcoding and not the anatomical connections between the stimuluslocations and the responding effectors.
Spatial compatibility effects also occur when responses are to beselected according to a non-spatial feature, such as the stimuluscolor or shape, but stimuli still vary in location. This phenomenon,called the Simon effect (Craft & Simon, 1970), is often attributedto automatic activation of the corresponding response that occurswhen the stimulus set has dimensional overlap, or similarity, withthe response set (Kornblum, Hasbroucq, & Osman, 1990). Forinstance, Wallace (1971) had participants make a left or rightkeypress to the shape of a stimulus in a left or right location, withuncrossed (left hand on left key and right hand on right key) orcrossed (left hand on right key and right hand on left key) handplacement. The Simon effect was the same size for the two handplacements, again implicating relative location coding.
For SRC tasks in which stimulus and response sets can berepresented in two dimensions, it has been shown that stimuli andresponses are coded as functions of their salient features: Fasterresponses occur when the more salient features of the stimulusand response sets correspond than when they do not (Proctor &Reeve, 1986). With arrays of four possible stimulus and responselocations, the salient feature is often the distinction between thetwo left and two right locations (e.g., Reeve & Proctor, 1984;Rubichi, Vu, Nicoletti, & Proctor, 2006), again, regardless of theanatomical mapping to responses.
This article was published Online First April 18, 2011.Dongbin (Tobin) Cho and Robert W. Proctor, Department of Psycho-
logical Sciences, Purdue University.We thank S. P. Tipper and M. A. Paul for providing us with the stimulus
files that they used in their 2006 study with A. E. Hayes and D. N. Bub andM. E. J. Masson for providing us with the stimulus files that they used intheir 2010 study.
Correspondence concerning this article should be sent to Dongbin (To-bin) Cho, Department of Psychological Sciences, Purdue University, 703Third Street, W. Lafayette, IN 47907-2081. E-mail: [email protected]
In addition, according to Hommel’s (1993b) referential codinghypothesis, spatial coding of stimulus location is in terms ofrelative position with respect to a reference object. In his experi-ments, the target stimulus in a Simon task was accompanied by anadditional irrelevant stimulus in the opposite location, whichserved as a referent object. The Simon effect was found in allexperiments. Moreover, Hommel and Lippa (1995) presented thetarget stimuli for a Simon task in the context of a face (left or righteye), rotated 90° to the left or right so that the stimulus locationswere up and down with respect to an egocentric axis. Regardlessof face orientation, the Simon effect was a function of correspon-dence of the left and right locations of the eyes in the face context.Considerable evidence thus indicates that spatial coding occurs interms of relative position with respect to the reference objects orframes.
Grasping Affordance
Although relative location coding is considered to be the majordeterminant of spatial SRC effects, relative location is only one ofmany object properties related to potential actions (Tucker & Ellis,1998). In contrast to the relative location view, Michaels (1988)argued that many compatibility effects may be based on utilizingthe affordances, or possibilities of action (Gibson, 1979), in S-Renvironments that contain relatively rich visual information.
Tucker and Ellis (1998) applied the concept of affordance tothe Simon effect using a task in which participants judged whetherthe orientation of a graspable object was upright or inverted. Theclassification was indicated by making a left or right keypress, inone experiment with the index fingers of each hand (between-handresponses) and in another with the index and middle fingers of theright hand (within-hand responses). For between-hand responses,these judgments of the orientation of a centrally placed graspableobject produced a Simon effect as a function of the left or rightlocation of the graspable part. However, for within-hand re-sponses, no significant Simon effect was found. Tucker and Ellisinterpreted these results as support for an affordance accountaccording to which the graspable part of the object activated agrasping response of the hand whose location corresponded to thatof the part.
Two aspects of Tucker and Ellis’s (1998) method and resultshave the potential to provide strong support for their affordanceaccount. First, the orientation judgments involved an object prop-erty relevant to grasping. If attending to a grasp-related property isnecessary to activate the corresponding grasping response, then theobject-based Simon effect should be found for other judgmentsthat involve an action-related property (e.g., object shape) but notfor judgments that involve properties unrelated to grasping (e.g.,object color). Tipper, Paul, and Hayes (2006) provided evidencefor this proposition, having participants make keypresses to theshape (a grasp-related property) or color of door-handle stimuliwith the handle to the left or right. They found an object-basedSimon effect as a function of door-handle direction for the judg-ments of object shape but not for the judgments of object color.Similarly, Loach, Frischen, Bruce, and Tsotsos (2008) found prim-ing effects of door-handle direction for judgments of texture butnot judgments of color. Both Tipper et al. and Loach et al.interpreted their results as indicating that a grasping affordance isnot activated automatically but only when a person’s task requires
attending to and discriminating an object property that is relevantto grasping.
The second potentially important aspect of Tucker and Ellis’s(1998) results is that the Simon effect was not significant when thetwo keypresses were made with index and middle fingers on thesame hand. Tucker and Ellis interpreted this lack of within-handSimon effect as converging evidence for their grasping affordanceaccount, because both responses on the same hand would beequally afforded or unafforded, depending on whether the grasp-able part was to the right or left. The crucial nature of this findingis made more evident by noting that studies examining the typicalSimon effect for stimulus location have found it to be at least aslarge for within-hand keypresses as for between-hand keypresses(Buckolz, O’Donnell, & McAuliffe, 1996; Heister, Ehrenstein, &Schroeder-Heister, 1987; Katz, 1981; Proctor & Vu, 2010). How-ever, as described in the next section, in a previous study usingfrying pan stimuli, we were unable to replicate the finding of anabsence of within-hand object-based Simon effect.
Relative Location Coding or Grasping Affordance?
Cho and Proctor (2010) conducted experiments to evaluate theobject property effect and lack of a within-hand object-basedSimon effect reported by Tucker and Ellis (1998). They examinedthe traditional Simon task, in which stimuli occur in left and rightlocations, and the object-based Simon effect. For tasks examiningthe latter, the relevant dimension was color in some conditions andupright/inverted orientation in others, and responses were madewith two fingers on the right hand or fingers on different hands. InExperiment 1, participants performed a location-based Simon taskwith the circle located to the left or right, or an object-based Simontask, with the object being a frying pan with the handle located tothe left or right, using either the within- or the between-handresponse set. For both tasks, the relevant stimulus dimension wascolor. Both within- and between-hand responses showed signifi-cant object-based Simon effects, and neither task showed a signif-icant difference in the Simon effect size within and between hands.Experiment 2 used the object-based Simon task with the relevantdimension being whether the frying pan was in upright or invertedorientation, as in Tucker and Ellis’s study. Two other groups ofparticipants made orientation judgments similar to the upright-inverted judgments of frying pan orientation, but for modifiedfrying pans with the tip of the handle separated from the pan andwith dashed lines at locations equivalent to the center of the longaxis of the panhandles. All stimulus sets showed significant Simoneffects, and none showed an absence or even reduction of theSimon effect for within-hand responses. Moreover, the color-judgment task of Experiment 1 produced a Simon effect as large asthat of the orientation-judgment task in Experiment 2. Cho andProctor concluded that relative location was the main factor intheir study contributing to the object-based Simon effect.
In addition to examining the Simon effects for mean RT andpercentage of error (PE), analyses of changes in the Simon effectacross the RT distributions are needed for detailed evaluation ofrelative location coding and grasping affordance. In Cho andProctor’s (2010) study, the Simon effect in the standard Simon taskcondition (the circle located to the left or right) decreased as RTincreased, a typical finding that is often attributed to initial rapidactivation of the corresponding response followed by dissipation
738 CHO AND PROCTOR
of that activation (e.g., Hommel, 1993a). In contrast, in the fryingpan conditions (a frying pan with the handle located to the left orright), the object-based Simon effect increased across the RTdistributions, suggesting that activation of the response corre-sponding to the handle took longer to occur. [See also Derbyshire,Ellis, and Tucker (2006); Phillips and Ward (2002), and Tuckerand Ellis (2001) for other examples of the increasing pattern of theobject-based Simon effect across the RT distributions.] However,this increasing function was evident in Cho and Proctor’s study forcolor judgments as well as orientation judgments, and for thewithin-hand response set as well as the between-hand response set,providing little evidence that it is due to processing of a graspingaffordance.
Consistent with a relative location coding account, in Cho andProctor’s (2010) frying pan study, the Simon effect was equallyevident (1) when the task involved a grasp-irrelevant color judg-ment as when it involved a grasp-relevant orientation judgment(i.e., there was no object property effect) and (2) when keypresseswere made with fingers on the same hand or on different hands(i.e., there was a within-hand Simon effect). Mean RTs and the RTdistributions suggested that the object-based Simon effect with thefrying pan stimuli occurs because the handle is spatially distinct,not because it is graspable. However, for the frying pan stimuli, thespatially distinct feature (relative location) and grasping feature
(grasping affordance) covaried with the direction of the handle.Consequently, if coding of relative location coding of the handlefeature were dominant, then this could have overridden any grasp-ing affordance associated with the handle, making the resultsappear as if there were no such affordance.
A stronger test of the grasping-affordance hypothesis can beperformed with stimuli for which the graspable handle is in aposition opposite that of the most distinct spatial feature. A teapot,one of the objects in Tucker and Ellis’s (1998) original study,satisfies this requirement because the graspable component, thehandle, is opposite that of the most distinct spatial feature, thespout. Therefore, in the present study, we conducted experimentsusing variations of teapot stimuli (see Figure 1). If the handleprovides a grasping affordance that opposes the correspondenceeffect due to the more spatially distinct spout, then specific pat-terns of the Simon effect should be evident that vary as a functionof whether the judgments involve (a) a property related to grasping(orientation) or not (color), (b) the responses are made with fingerson different hands or the same hand, and (c) the stimuli are intactteapots with both spout and handle features or altered teapots withonly a spout or handle.
In Experiment 1, participants responded to the teapot stimuliwith between- and within-hand response sets in one of two con-ditions, color judgments, or upright-inverted orientation judg-
Figure 1. Examples of the stimuli used in Experiments 1–3. The stimuli were based on a teapot from the Webpage http://math.hws.edu/bridgeman/courses/324/s06/doc/tutorial/helloteapot.jpg.
739CORRESPONDENCE EFFECTS FOR OBJECTS WITH PROTRUSION
ments. The Simon effect in general often reflects the combinedinfluences of correspondence with respect to different componentsor frames of reference, weighted according to their relative sa-lience (e.g., Lamberts, Tavernier, & d’Ydewalle, 1992). Conse-quently, we expected the Simon effect to be determined by corre-spondence of the more salient spatial feature, the spout, with thelocation of the correct response to the relevant stimulus property.The handle should contribute an opposing component, renderingthe Simon effect relatively small.
If the contribution of the handle is due in whole or part to itsproviding a grasping affordance, then the Simon effect in rela-tion to the spout should be smaller with the between-handresponse set, for which this opposing affordance is a factor,than for the within-hand response set, for which it is not. Sucha result could also be obtained, though, if the contribution of thehandle component is due to its relative location, because thereis evidence to suggest that response competition is stronger fortwo fingers within the same hand than for two fingers ondifferent hands (e.g., Kornblum, 1965; Reeve & Proctor, 1988).Buckholz et al. (1996) appealed to such difference in competi-tion to explain why the location-based Simon effect was largerwithin than between hands in their study. Assuming, consistentwith Tipper et al. (2006) and Loach et al. (2008), that a graspingaffordance is activated only for judgments related to a graspableobject property, the affordance account predicts that the patternof larger within- than between-hand Simon effect should befound only for the upright/inverted object judgments. In con-trast, the location coding account predicts that this patternshould also be evident for the color judgments.
For Experiment 2, the handle was removed from the originalteapot in order to examine the object-based Simon effect whenonly the most spatially distinct feature, the spout, was present. Ifthe Simon effect in Experiment 1 was a function of the combinedeffects of the spout and handle, then the Simon effect relative tothe spout location should be larger in Experiment 2 for which thereis no opposing handle feature. An absence of difference in size ofthe Simon effect for the within- and between-hand response setswould suggest that the difference evident with the intact teapot isdue to greater competition for within-hand responses when eachresponse is activated by a different feature.
For Experiment 3, the spout was removed from the originalteapot, leaving the handle as the only attached component. In thiscase, the object-based Simon effect should be determined by thehandle. If the handle affords grasping, then the Simon effect withreference to the handle should be larger between-hands thanwithin-hands, whereas if the handle’s effect is due to its relativelocation, then there should be no difference in Simon effect size forthe between- and within-hand response sets.
Finally, because Experiments 1–3 showed an object-based Si-mon effect for color judgments, whereas Tipper et al.’s (2006)study with door-handle stimuli did not, in Experiment 4 we ex-amined whether the specific way we displayed our stimuli wasresponsible for the different results. Tipper et al.’s door-handlestimuli were used for two display conditions, one comparable toours in which the base of the door handle was centered, as were theobjects in our other experiments, and the other for which thehandle was centered, as in their study.
Experiment 1
Experiment 1 was designed to obtain evidence regarding howthe two protruding components of the teapot stimulus (spout andhandle) contribute to the object-based Simon effect and whetherthis is in a manner consistent with that for the frying pan stimulusthat had only one protruding component. Because the intact fryingpan experiments in Cho and Proctor’s (2010) study suggested thatrelative location coding was the main factor determining theobject-based Simon effect, the overall tendency of the resultsshould be toward a positive object-based Simon effect with respectto spout position, because the spout is more distinctive spatiallythan the handle. In contrast, if Tucker and Ellis’s (1998) absenceof within-hand Simon effect is still valid, the within-hand Simoneffect should shift to an even more positive value due to locationcoding of the spout, which is the main contributor to the within-hand Simon effect.
As noted, participants were assigned to one of two judgmentconditions: color judgment and upright-inverted orientation judg-ment. If the Simon effect relative to the spout were smallerbetween than within hands but for only the upright/inverted tea-pots, which would result in a three-way interaction of Correspon-dence � Response Mode � Condition (green/red teapot or up-right/inverted teapot), the evidence would favor a graspingaffordance. If the three-way interaction were not apparent, thiswould provide evidence suggestive of relative location coding.This is because the color condition would show similar patterns asthe orientation condition, which would imply that action relevanceof the judgment is not a critical factor for the object-based Simoneffect.
Method
Participants. Sixty-four students from Introductory Psychol-ogy classes at Purdue University participated for credits toward acourse requirement. All participants had normal or corrected-normal vision and normal hearing, and they were naive as to thepurpose of the experiment. Participants were randomly assigned toone of four groups determined by response mode (within- orbetween-hand) and stimuli (green/red teapots or upright/invertedteapots), for a total of 16 participants in each group.
Stimuli and apparatus. Stimuli were pictures and tonesgenerated by Micro Experimental Laboratory (MEL 2.01), whichcontrolled the experiment. For all conditions, the participant satdirectly in front of the monitor, at a distance of approximately 60cm. The stimuli were teapots (5.3 cm height; 12 cm width, seeFigure 1) displayed with the pot centered on the screen, but withthe location of the handle (2 cm in length) being to the left or right.For these teapots, the length of the spout (4 cm) is approximatelytwice that of the handle (2 cm), areas of approximately 3.83 and2.78 cm2, making the spout more spatially distinctive. To verifythis difference in distinctiveness, 34 different undergraduate stu-dents from the same subject pool were shown a gray picture of theteapot against a black background (handle on the left for half of thestudents and on the right for the other half) and asked to indicatewhich component seemed most prominent. Twenty-four studentsselected the spout and only 10 selected the handle, �2 � 5.77, p �.016.
For the main experiment, the teapot stimuli were green or red(MEL color code � 2 for green and � 4 for red) for color
740 CHO AND PROCTOR
judgment tasks and gray (MEL color code 7) for orientationjudgment task. The handle was located 90° from the center verticalline of the stimuli to either side. Responses were registered bypresses of one or the other of two adjacent keys on the bottom rowof the computer keyboard (the B or N key) with the appropriatefingers. With each stimulus set, half of the participants respondedwith the index fingers of the left and right hands, whereas halfresponded with the index and middle fingers of the right hand.
Procedure. Each condition consisted of two blocks of 176trials in which each object occurred equally often in each color (ororientation) and with a left or right handle location, with orderrandomized for each participant. Participants were told to make aleft or right response depending on whether the object was greenor red color or upright/inverted orientation. The first trial blockwas performed with one S-R mapping (e.g., green-left, red-right)and the second block with the other (e.g., red-left, green-right),with the mapping order counterbalanced across participants.1 Par-ticipants were told to respond as fast and accurately as possible,without making too many errors. Each participant received 16practice trials before each block.
Each trial began with onset of the blank screen for 1,000 ms;then a stimulus appeared and was presented until a response wasmade or for 1,500 ms, at which time the trial was terminated if noresponse had been made. Participants were not given feedback onresponse latencies, but errors were immediately followed by a400-Hz short tone (500 ms) from the computer, followed by onsetof the next trial.
Results
Mean RT and PE. Mean RT for correct responses and PEwere computed for each participant (see Table 1). Trials withRT � 200 ms or � 1,500 ms were discarded from the subsequentanalysis (� 1.0% of all trials in this and the other experiments). A2 (correspondence relative to the spout: corresponding, non-corresponding) � 2 (response mode: between hand, withinhand) � 2 (condition: green/red teapot, upright/inverted teapot)ANOVA was conducted for RT and PE, with repeated measures onthe first factor.
The overall mean RT and PE were 490 ms and 2.2%, respec-tively. RT differed significantly between the two stimulus condi-tions, F(1, 60) � 7.48, MSe � 8,027, p � .008, �p
2 � .11, but errorrate did not, F(1, 60) � 2.57, MSe � 8.52, p � .11. RT was longerin the upright/inverted teapot condition (M � 512 ms) than in thegreen/red teapot condition (M � 469 ms). Responses tended to beslower for the within-hand mode (M � 503 ms) than for thebetween-hand mode (M � 477 ms), but this difference was notsignificant, F(1, 60) � 2.64, MSe � 8,027, p � .11 (for PE, F �1), and it did not interact with condition for RT or PE, Fs � 1.
Both RT and PE showed main effects of correspondence, Fs(1,60) � 96.48 and 32.87, MSes � 99.21 and 1.79, ps � .001, �p
2 �.62 and .35, respectively, yielding Simon effects of 17 ms and1.3% (i.e., responses were faster and more accurate when the spoutand response corresponded than when they did not). Correspon-dence and response mode interacted significantly for RT andalmost so for PE, Fs(1, 60) � 9.84 and 3.26, MSes � 99.21 and1.79, ps � .003 and .076, �p
2 � .14 and .05. The Simon effect wasgreater for within-hand responses (23 ms; 1.8%) than for between-hand responses (11 ms; 1.0%). The Correspondence � Condition
interaction was not significant for RT, F � 1.0, but approachedsignificance for PE, F(1, 60) � 3.46, MSe � 1.79, p � .068, �p
2 �.05. The PE Simon effect tended to be larger for the upright/inverted condition (1.8%) than for the green-red teapot condition(0.9%). Of importance, there was no sign of a three-way interac-tion of correspondence, response mode, and condition for RT orPE, Fs(1, 60) � 1, indicating that the pattern of 12-ms greaterwithin-hand than between-hand Simon effect did not vary reliablyacross the red/green teapot and upright/inverted teapot conditions.
A supplementary analysis on the RT data for the upright/inverted teapot condition was conducted that distinguished theupright orientation, which would be more natural and might morestrongly afford grasping, from the inverted orientation. Correspon-dence interacted with orientation, F(1, 30) � 7.79, MSe � 352,p � .009, �p
2 � .21, showing a larger Simon effect for the uprightteapot (28 ms) than for the inverted teapot (10 ms). Note that thislarger effect for the upright teapot was relative to the spout ratherthan the handle, counter to the possibility that the grasping affor-dance is stronger for the more natural orientation. The Simoneffect was larger for the within-hand mode than the between-handmode, as in the overall ANOVA, although the interaction ofcorrespondence with response mode did not quite attain statisticalsignificance, F(1, 30) � 3.38, MSe � 194, p � .076, �p
2 � .10.However, correspondence and response mode did not enter into athree-way interaction with orientation, F � 1.0, indicating that thistendency for larger within- than between-hand Simon effects wassimilarly evident for the upright and inverted teapots (being 12 and7 ms, respectively). We will defer discussion of this finding untilthe General Discussion.
RT distribution analyses. The Simon effect across the RTdistributions was determined as follows. For all conditions, RTsfor corresponding and non-corresponding conditions were rankordered for each participant and equally divided into four bins inwhich Simon effects were calculated. An ANOVA with the fourbins, two conditions, and two response modes as factors wasperformed. This ANOVA showed a main effect of bin and atwo-way interaction of bin with response mode, Fs(3, 180) �21.74 and 9.49, MSe � 365, ps � .001, �p
2 � .27 and .14. TheSimon effect increased across RT bins, with the increase at the lasttwo bins being more pronounced for the within-hand responsesthan for the between-hand responses (see Figure 2). However, nothree-way interaction of bin, response mode, and condition wasfound, F(3, 180) � 1.84, MSe � 365, p � .14, indicating that thispattern did not differ reliably across the two conditions.
1 To make sure that changing the mapping of the relevant dimension toresponse did not influence the patterns of results, we conducted analysesfor Experiments 1, 2, and 3 that included block (first mapping vs. secondmapping) as a factor. The RT data showed an interaction of Correspon-dence � Block that was not significant in Experiment 1, F(1, 60) � 2.46,MSe � 217, p � .12, �p
2 � .04, but was in Experiments 2 and 3, Fs(1, 60) �13.21 and 14.11, MSes � 337 and 150, ps � .001, �p
2 � .18 and .19. In allcases, the Simon effect was smaller in the second block than in the first. Nohigher order interactions were significant, except for the three-way inter-action of Correspondence � Block � Condition in Experiment 3, F(1,60) � 4.97, MSe � 150, p � .029, �p
2 � .077, which reflected only a largerdifference in the Simon effect for the first second blocks for the orientationjudgments (Simon effects of 26 and 8 ms) than for the color judgments(Simon effects of 19 and 14 ms).
741CORRESPONDENCE EFFECTS FOR OBJECTS WITH PROTRUSION
Discussion
Consistent with the relative location coding hypothesis, Simoneffects were obtained with regard to the most spatially salient partof the object, the spout. For the orientation judgments, whichinvolve a grasp-related property, the within-hand Simon effect waslarger than the between-hand Simon effect, suggesting that thehandle had a bigger effect for between-hand responses than within-hand responses. This result could be due to the handle affordinggrasping because the affordance would be absent for within-handresponses. However, color judgments, which do not involve agrasp-related property, also showed a larger within-hand thanbetween-hand Simon effect, and the three-way interaction of cor-respondence, response mode, and condition did not approach sta-tistical significance. Thus, if a grasping affordance is involved, itwould have to have been activated regardless of whether the taskrequired judging a grasp-related feature, counter to the conclusionsof Tipper et al. (2006) and Loach et al. (2008).
Bub and Masson (2010) recently reported results of an experi-ment in which between-hand keypress responses were made toteapot stimuli, and they found no Simon effect. Because theteapots were symmetric, they were displayed with the pot centeredon the screen, as in our experiment. The most likely reason why weobtained a 10-ms Simon effect for between-hand responses andthey found no effect is that the spout was longer and more salientthan the handle for our stimuli but not for theirs.
The RT distribution analyses for Experiment 1 showed anoverall increasing Simon effect with regard to the location of thespout across the four bins, as has been reported previously forobject-based Simon effects (e.g., Phillips & Ward, 2002). How-ever, different patterns of RT distributions for within- andbetween-hand responses were found. For within-hand responses,the Simon effect increased monotonically across the distribution inboth conditions, whereas the increase was less for between-handresponses, with the mean Simon effect decreasing slightly at thelast RT bin (see Figure 2). This decrease suggests that, for thebetween-hand responses, activation of the code corresponding tothe more salient spout location either dissipated or was countered
later in time by opposing activation of the code corresponding tothe less salient handle. An implication of the distribution analysesis that the larger Simon effect for the within-hand response modeoccurred mainly for responses with longer latencies.
As stated before, there was no three-way interaction of corre-spondence, response mode, and condition for RT or PE, indicatingthat the patterns of within- and between-hand Simon effects weresimilar for the two conditions. Moreover, there was no three-wayinteraction of RT bin, response mode, and condition for the RTdistribution analysis, either. These data show that the orientationjudgments did not yield results that differed reliably from those ofthe color judgments with regard to the object-based Simon effectfor keypress responses. Similar results were obtained in Cho andProctor’s (2010) frying pan experiments, where color and orien-tation judgments yielded comparable object-based Simon effects.
Experiment 2
The larger object-based Simon effect for within-hand responsesthan for between-hand responses in Experiment 1 may have oc-curred because each stimulus contained two components in oppo-site locations, possibly exacerbating problems associated withgreater response competition for within-hand keypress responses(e.g., Buckholz et al., 1996). This reasoning suggests that an objectwith only one location component should not generate a largerwithin- than between-hand Simon effect. To test this hypothesis,for Experiment 2, a teapot with only the most spatially salientlocation component (the spout) was constructed by removing thehandle, and the within- and between-hand Simon effects werecompared. The handle was removed for this experiment, ratherthan the spout, because the Simon effect was obtained relative tothe spout position rather than the handle position in Experiment 1.
Therefore, in Experiment 2, participants in within- and between-hand conditions performed a color-judgment or orientation-judgment task with the handle removed from the teapot, leavingthe most spatially salient part, the spout, intact. Based on our priorresults with the frying pan stimuli (Cho & Proctor, 2010), weexpected a large Simon effect that would not differ in size for the
Table 1Mean Response Times (RTs) and Percentage Errors (PEs) as Functions of Correspondence and Response Mode, and theSimon Effect in Experiments 1–4
within- and between-hand response conditions. Such results wouldbe consistent with the hypothesis that the handle provided acorrespondence effect in Experiment 1 that opposed that producedby the spout location. This pattern would also be consistent with anaffordance account, on the assumption that the spout is not per-ceived as affording grasping, even though it is now the soleappendage. If the spout does afford grasping in this case, then theSimon effect should be larger for the between-hand response mode(for which the affordance distinguishes a grasp with the left orright hand) than for the within-hand response mode (for which itdoes not).
Method
Sixty-four new students participated, 32 each with the red/greenteapots and with the upright/inverted teapots; within each of thosegroups, 16 used the between-hand response set and 16 the within-hand response set. All procedures were identical to Experiment 1,except for the stimuli. Stimuli were teapots with spout only and no
handle (spout: 4 cm long, located approximately 50° from thecenter vertical line of the stimuli; see Figure 1), displayed with thepot centered on the screen.
Results
Mean RT and PE. The mean RT and PE data are shown inTable 1. The overall mean RT and PE were 489 ms and 2.05%. RTwas longer and PE greater for the upright/inverted teapot condition(Ms � 516 ms and 2.6%) than in the green/red teapot condition(Ms � 463 ms and 1.5%), Fs(1, 60) � 8.93 and 4.72, MSes �10,147 and 8.05, ps � .004 and .034, �p
2 � .13 and .07. Neithermeasure showed a main effect of response mode nor an interactionof response mode with condition, all of the F ratios being less than1 except that of the response-mode main effect for PE, F(1, 60) �3.52, p � .065.
Both RT and PE showed main effects of correspondence, Fs(1,60) � 141.30 and 40.65, MSes � 202 and 2.30, ps � .001, �p
2 �.70 and .40, respectively, yielding Simon effects of 30 ms and
Figure 2. Between- and within-hand Simon effects plotted as a function of the mean RT for each quartile inthe different task conditions of Experiments 1–3.
743CORRESPONDENCE EFFECTS FOR OBJECTS WITH PROTRUSION
1.7%. These effects did not interact with condition, Fs(1, 60) �0.37 and 3.26, ps � .448 and .076, although there was a tendencyfor the PE Simon effect to be larger for the upright-invertedjudgments (2.2%) than for the color judgments (1.2%). Mostimportant, correspondence and response mode did not interact foreither RT or PE, Fs(1, 60) � 1.0, nor did they enter into athree-way interaction with condition, Fs(1, 60) � 1.79 and 1.84,ps � .18. Thus, the Simon effect did not differ between the tworesponse modes, and this result generalized across both the up-right/inverted and color judgments.
Unlike Experiment 1, the supplementary analysis on the RT datafor the upright/inverted teapot condition showed no 2-way inter-actions of correspondence with upright/inverted orientation orresponse mode, Fs(1, 30) � 1.75, ps � .19. The three-way inter-action of these variables also was non-significant, F(1, 30) � 1.22,p � .78. The Simon effect was of similar size for the upright andinverted spout-only teapots (31 and 25 ms).
RT distribution analysis. An ANOVA of the RT distribu-tions similar to that of Experiment 1 showed a main effect of RTbin, F(3, 180) � 40.42, MSe � 448, p � .001, �p
2 � .40. RT bindid not interact with response mode or condition separately, F �1, or in combination, F(3, 180) � 2.59, p � .095 (with Huynh-Feldt adjustment). Thus, both within- and between-hand conditionsshowed an increase in the Simon effect across RT bins, Fs(3, 90) �35.15 and 12.06, MSes � 334 and 562, ps � .001, �p
2 � .54 and .29,respectively (see Figure 2).
Discussion
Unlike Experiment 1, the Simon effect for both RT and PE didnot differ significantly between the within-hand and between-handresponse sets. RT distribution analyses for Experiment 2 showedan overall increasing Simon effect with regard to the location ofthe spout across the four bins. No significant differences in thepattern of RT distributions for within-hand and between-handresponses were found: For both response modes, the Simon effectincreased across the distribution. Thus, both mean RT and RTdistribution data suggest that within- and between-hand Simoneffects are similar.
The overall Simon effect of 30 ms in Experiment 2 was largerthan that of 18 ms in Experiment 1, F(1, 124) � 16.81, MSe � 150,p � .001, �p
2 � .12. This result supports the view that the reasonwhy the Simon effect was relatively small in Experiment 1 is thatthe teapot handle opposed the relative location of the spout. Thisdifference in Simon effect sizes for Experiments 1 and 2 did notinteract with response mode, F(1, 124) � 1.39, p � .24, indicatingthat the influence of the handle did not depend on whether theresponses were from the same or different hands. The resultsobtained in Experiment 2, for which the teapot had only a singleappendage, are in close agreement with those of Cho and Proctor’s(2010) frying pan stimuli, which also had only a single appendageand showed Simon effects of similar size for the between- andwithin-hand response sets. Because there was no sign of the spoutaffording grasping in Experiment 2, when it was the only append-age, the Simon effects are most clearly attributable to spatialcoding. Because the appendage in Cho and Proctor’s frying panstimuli unambiguously afforded grasping, and yet still showed nolarger Simon effect between-than within-hands, the most straight-
forward account is that grasping affordances are playing little ifany role.
Experiment 3
Experiment 3 was designed to be the counterpart to Experiment2 by using a teapot with the spout removed and only the graspablehandle remaining. The handle should produce a Simon effect basedon its relative location, but because it is less salient than the spout,the Simon effect should be smaller than that produced by the spoutin Experiment 2. If the smaller Simon effect in Experiment 1 thanin Experiment 2 were due to the handle producing an opposingeffect that countered that of the spout, then the Simon effectproduced by the handle in Experiment 3 should be approximatelythe same as the difference in size of Simon effects for Experiments1 and 2. We hypothesized earlier that the larger within-hand thanbetween-hand Simon effect in Experiment 1 was due to the teapotstimuli having opposing components. If so, the within- andbetween-hand Simon effects relative to the handle location inExperiment 3 should be of similar size, because the handle is thesole component.
That the appendage was a handle allows another opportunity toobtain evidence of a grasping affordance. Following the logic ofTucker and Ellis (1998), a contribution of grasping affordancewould be shown by a larger between- than within-hand Simoneffect. If a grasping affordance is not activated when making colorjudgments, as Tipper et al. (2006) and Loach et al. (2008) con-cluded to be the case for door-handle stimuli, then a Simon effectshould occur only for orientation judgments made with thebetween-hand response set.
Method
Sixty-four students, who did not participate in Experiment 1 or2, participated for credits toward a course requirement. All proce-dures in this experiment were identical to Experiment 2 except thestimuli. The stimuli were teapots with only handles and no spouts(see Figure 1) displayed with the pot centered on the screen and thehandle location being to the left or right.
Results
Mean RT and PE. The mean RT and PE data are shown inTable 1. The overall RT and PE were 472 ms and 2.04%. RT waslonger and PE greater for the upright/inverted teapot condition(Ms � 516 ms and 2.6%) than in the green/red teapot condition(Ms � 463 ms and 1.5%), Fs(1, 60) � 10.85 and 5.07, MSes �7,822 and 6.75, ps � .002 and .028, �p
2 � .15 and .08. Neithermeasure showed a main effect of response mode or an interactionof response mode with condition, Fs � 1.
Both RT and PE showed main effects of correspondence, Fs(1,60) � 68.90 and 14.82, MSes � 128 and 1.62, ps � .001, �p
2 � .53and .20, respectively, yielding Simon effects of 16 ms and 0.86 %.Separate analyses showed these effects were reliably smaller forthe handle in Experiment 3 (16 ms and 0.6%) than for the spout inExperiment 2 (30 ms and 1.7%), Fs(1, 120) � 17.04 and 5.80,MSes � 165 and 1.96, ps � .001 and .018, �p
2 � .124 and .046. ForExperiment 3, neither RT nor PE showed a main effect of responsemode, Fs(1, 60) � 1.36, p � .25, respectively. These effects did
744 CHO AND PROCTOR
not interact significantly with condition or with response mode foreither RT or PE, Fs(1,60) � 1.0. The F ratio for the three-wayinteraction was also less than 1.0 for RT, although for PE itapproached the .05 level, F(1, 60) � 3.82, p � .055. Although thePE Simon effect tended to be larger for between- than within-handresponses for upright/inverted judgments, this tendency was notsignificant, F(1, 30) � 2.44, p � .129, nor was the oppositetendency for color judgments, F(1, 30) � 1.39, p � .248.
Although the Simon effect was obtained relative to the handle inthis experiment, compared to relative to the spout in Experiment 1,RTs for the upright/inverted teapot condition showed an interac-tion of correspondence with orientation, F(1, 30) � 4.86, MSe �180, p � .035, �p
2 � .14, but no three-way interaction withresponse mode, F(1, 30) � 1.44, p � .24. The Simon effectrelative to the handle was larger for the upright teapot (23 ms) thanfor the inverted teapot (12 ms). Additionally, there was no inter-action of correspondence with response mode, F � 1.0.
RT distribution analysis. An ANOVA of the RT distribu-tions similar to that of Experiments 1 and 2 showed a maineffect of RT bin, F(3, 180) � 30.28, MSe � 341.84, p � .001,�p
2 � .39. RT bin did not interact with response mode orcondition separately, Fs � 1, or in combination, F(1, 180) �1.97, p � .12. Thus, like the RT distribution results in Experiment2, both within- and between-hand conditions showed the pattern ofincreasing Simon effects across the RT bins, Fs(3, 90) � 20.93 and17.22, MSes � 379 and 305, ps � .001, �p
2 � .41 and .37, respectively(see Figure 2).
Discussion
A Simon effect for handle location was obtained that did notinteract significantly with response mode, results in agreementwith those obtained with frying pan stimuli by Cho and Proctor(2010) and with the relative location coding account. Given thatthe upright-inverted judgment task, which was used in Tucker andEllis’s (1998) study, did not show a larger between-hand Simoneffect or absence of within-hand Simon effect, it is reasonable toconclude that the Simon effect is due to relative location of thehandle.
Consistent with the view that the Simon effect in Experiment 1was determined by the smaller correspondence effect of the handleopposing the larger correspondence effect of the spout, the Simoneffect for the handle alone in this experiment was significantlysmaller than the Simon effect for the spout alone in Experiment 2.The 16-ms size was similar to the difference in Simon effect forthe spout-only teapot in Experiment 2 and the intact teapot inExperiment 1 (15 ms). Thus, the overall size of the Simon effect inExperiment 1 is predicted well by the opposing effects of the spoutand handle, although the larger effect size within-hands thanbetween-hands was evident only in Experiment 1 in which the twocomponents were present in opposition.
RT distribution analyses showed that, as in Experiment 2, theSimon effect was an increasing function of RT for both within-and between-hand responses. This finding also suggests thatthere is little difference in underlying mechanisms betweenwithin- and between-hand Simon effects when the stimulus hasonly one location component.
Experiment 4
Experiments 1–3 and those with frying pan stimuli reported inCho and Proctor (2010) have consistently shown object-basedSimon effects for RT with color judgments that are of similar sizeas those obtained with orientation judgments. These results are incontrast to those of Tipper et al. (2006), who found a Simon effectfor door-handle stimuli when the task required shape judgmentsbut not when it required color judgments. To reconcile this dis-crepancy, we conducted Experiment 4 using Tipper et al.’s door-handle stimuli, with two display conditions. For one, the base ofeach handle was centered, similar to the way we presented ourstimuli, whereas for the other, the handles were centered. Onlybetween-hand responses (the left and right index fingers) wereused.
Method
Thirty-two students, who had not participated in the previousexperiments, participated for course credits. Half of the partici-pants were assigned to a base-centered condition and half to ahandle-centered condition. For both conditions, the stimuli werethose used by Tipper et al. (2006), with the handle location beingto the left or right (see Figure 3)2, and the instructions were wordedidentically to theirs. The stimuli were the same size as in theirstudy, being generated from their files on a 19-in. monitor(1,280 � 1,024 resolution), as in their study. The lengths of thesquare and round door handles were 9.5 cm and 10 cm, respec-tively, with the base attachment being 3.5-cm diameter. For thebase-centered condition, the base for each door handle was dis-played at the center of the screen, with the handle to one side or theother, whereas for the handle-centered condition, the handle wascentered on the screen, with the base varying in position. Theprocedures were otherwise similar to those of Experiments 1–3,except that participants had the same S-R mapping for the relevantdimension across the trial blocks and were required to take a 1-minbreak after the first block, as in Tipper et al.’s study.
Results
We analyzed the data as a function of correspondence (corre-sponding, non-corresponding), and display condition (base-centered, handle-centered). The mean RT and PE data as a func-tion of those variables are shown in Table 2. The overall error ratewas 1.9%. Both RT and PE showed main effects of correspon-dence, Fs(1, 30) � 11.83 and 4.15, MSes � 197 and 4.55, ps �.002 and � .05, �p
2 � .28 and .12, respectively, yielding Simoneffects of 12 ms and 1.1 %. Correspondence and display conditioninteracted for both RT, F(1, 30) � 36.88, p � .001, �p
2 � .55, andPE, F(1, 30) � 13.73, p � .001, �p
2 � .31. The Simon effect wassignificant for the base-centered condition (33 ms; 3.0%) but notfor the handle-centered condition (�9 ms; �0.8%).
2 The active vs. passive distinction shown in Figure 3 was inadvertentlynot coded for the base-centered condition. It is not of central importance toour determination of whether a Simon effect occurs with color judgments,for which neither stimulus type yielded a Simon effect in Tipper et al.’s(2006) study.
745CORRESPONDENCE EFFECTS FOR OBJECTS WITH PROTRUSION
Discussion
This experiment showed an object-based Simon effect for colorjudgments with the door-handle stimuli when they were positionedas the stimuli in our Experiments 1–3 were, with the base of thehandle centered. However, when the door-handle stimuli werepositioned as in Tipper et al.’s (2006) study, with the handlecentered, there was no Simon effect relative to handle location, anda tendency for a reverse Simon effect (i.e., one based on corre-spondence of the base location with the response). The likelyreason why Tipper et al. obtained a positive Simon effect whenparticipants had to respond based on the shape of the handle is thatthis judgment caused them to code the handle relative to the base,even though it was the latter that was physically changing left andright sides.
In Bub and Masson’s (2010) recent experiment, mentionedearlier, they also found no object-based Simon effect for the handlelocations of beer mugs when participants responded to the colorsof the mugs with between-hand keypresses. The beer mugs werepresented such that the entire object, including the handle, wascentered on the screen, rather than with the body centered. There-fore, their results are consistent with those of Tipper et al. (2006)and the handle-centered condition in the present Experiment 4 inshowing no Simon effect.
General Discussion
For the variations of teapot stimuli used in Experiments 1–3, theSimon effect was evident in the mean RT data for all conditionsregardless of whether the keypress responses were within or be-tween hands. Averaged across the within- and between-hand re-sponse modes, the overall Simon effect ranged from 17 ms for thehandle-only teapot stimuli in Experiment 3 to 30 ms for thespout-only stimuli in Experiment 2. The smallest effects acrossthe conditions were obtained when the more salient spout wascountered by correspondence with regard to the handle (Experi-ment 1) and when the less salient handle was the only componentcontributing to the Simon effect (Experiment 3).
The overall Simon effects with regard to the spout in thecolor-judgment conditions of Experiments 1 and 2 were 17 and32 ms, respectively, a difference of 15 ms. For Experiment 3,the Simon effect with regard to the handle in the color-judgmentcondition averaged 16 ms. Thus, the Simon effect for thecomplete teapot in Experiment 1 was approximately equal tothat for the spout-alone teapot in Experiment 2 minus theopposing effect for the handle-alone teapot in Experiment 3.The relative sizes of the Simon effects for the spout and handlealso match up to the ratio of their physical lengths from the potto the farthest extreme of the part (spout: 32 ms and 4 cm;
Figure 3. The right-facing door-handle stimuli used in Experiment 4. The left-facing stimuli were similar butwith the handle to the left side. Adapted from “Vision for Action: The Effects of Object Property Discriminationand Action State on Affordance Compatibility Effects,” by S. P. Tipper, M. A. Paul, and A. E. Hayes, 2006,Psychonomic Bulletin & Review, 13(3), 493–498, with the author’s permission.
Table 2Mean Response Times (RTs) and Percentage Errors (PEs) as Functions of Correspondence and Response Mode, and theSimon Effect in Experiment 4 and Tipper et al.’s Original Data
Handle-centered 417 (69.71) 1.6 (1.62) 408 (67.01) 0.8 (0.54) �9 �0.8Tipper et al.’s (2006) original study 487 2.2 486 3.4 �1 1.2
� p � .05.
746 CHO AND PROCTOR
handle: 16 ms and 2 cm; see Figure 1), emphasizing the relativesalience of the components.
The results for the orientation judgments generally followed thispattern, although not quite as clearly as those for the color judg-ments (Simon effects of 19, 29, and 17 ms in Experiments 1, 2, and3, respectively). When separated into upright versus inverted stim-uli, the inverted stimuli showed a pattern more similar to that of thecolor-judgment condition (Simon effects of 10, 25, and 12 ms inExperiments 1, 2, and 3, respectively) than did the upright stimuli(Simon effects of 28, 31, and 22 ms in Experiments 1, 2, and 3,respectively). Although the Simon effect was larger for uprightthan inverted stimuli, F(1, 90) � 13.07, MSe � 497, p � .001, �p
2
� .127, the interaction of orientation with experiment was notsignificant, F(2, 90) � 1.31, p � .28. These data suggest that thejudgments of “upright” may be based more on the directionalcharacteristics of the object as a whole (i.e., as pointing to the leftor right), with the Simon effect being slightly smaller when handleis the determining factor because of its lesser salience.
The relative locations of both protruding components, the spoutand handle, contributed to the Simon effect. One of the compo-nents, the handle, could possibly provide a grasping affordance.However, in all conditions across Experiments 1–3, the within-hand response set, for which the grasping affordance should notfavor one or the other response, showed significant Simon effects.These effects were at least as evident when the task requiredjudgments about color (a property unrelated to grasping) ratherthan orientation (a property related to grasping). These resultsprovide little evidence that the observed object-based Simon ef-fects were due to a grasping affordance. Note also that the similarsize of Simon effects in the orientation-judgment condition forupright intact teapots (28 ms) and spout-only teapots (31 ms)suggests that a grasping affordance is not involved because thehandle on the intact teapot did little to counter the directionalcomponent of the object.
The results appear to confirm that relative location coding is themain contributor to the object-based Simon effect. Unlike previousstudies, the within-hand Simon effect was larger than the between-hand Simon effect for color judgments in Experiment 1. Thispattern of results was evident for the combined results of green/redteapot and upright/inverted teapot conditions. Moreover, there wasno significant three-way interaction of Correspondence � Re-sponse Mode � Condition, which shows that the result pattern(within-hand Simon effect being greater than between-hand Simoneffect) was similar across the conditions in Experiment 1. With thehandle or spout removed (Experiments 2 and 3), the size of thewithin-hand Simon effect did not differ reliably from that of thebetween-hand Simon effect.
Why was the within-hand Simon effect larger than the between-hand Simon effect with the intact teapot? We speculated that thisdifference was caused by having two location components inopposite directions. When the handle was removed in Experiment2, there was no significant difference between the within- andbetween-hand response modes, although the means of the upright/inverted condition tended to show a larger within-hand Simoneffect. Moreover, because the overall Simon effect was twice aslarge as in Experiment 1, one could argue that elimination of thedifference was a consequence of increasing the size of the Simoneffect. The results of Experiment 3 rule out this possibility, though,because there was no difference in Simon effect sizes for the two
response modes when the teapot had only a handle, which yieldedan overall Simon effect of similar size to that in Experiment 1.Moreover, only with the intact teapot in Experiment 1 did the RTdistributions for the between-hand Simon effect show a down-turnat the longest RT bin rather than increasing monotonically acrossthe distribution. These results are consistent with the hypothesisthat the larger within-hand than between-hand Simon effect inExperiment 1 is due to the stimuli having location components thatare in opposition.
Kornblum (1965) provided evidence that within-hand responsesare more difficult (i.e., yield longer RT) than between-hand re-sponses to stimuli that vary along a single dimension. He inter-preted this finding as indicating that response competition isgreater for within- than between-hand responses. Buckholz et al.(1996) attributed a larger within- than between-hand Simon effectthat they obtained to greater competition between two responseson the same hand than two responses on different hands. If thereis a tendency toward more competition when the two responsealternatives are on the same hand, then having opposing locationcomponents may exacerbate this tendency by producing compet-ing activation of both fingers on each trial.
The results support the importance of relative location coding inthe object-based Simon task, a point that was also stressed byBosbach, Prinz, and Kerzel (2005) and Cho and Proctor (2010).Bosbach et al. investigated the effect of task-irrelevant motioninformation on left-right responses. They concluded that ratherthan the effect being a result of the motion being processed asdirectional information, it is a result of participants coding shifts inposition relative to the beginning position. Cho and Proctor alsoinvestigated the task-irrelevant color and orientation informationon left-right responses and concluded that relative location codingis the entire factor for frying pan stimuli that have only one spatialattachment. The results of the present experiments are consistentwith those studies because when the irrelevant dimension was leftor right location of spatially distinct object features, in the case ofExperiments 1–3, the spout and the handle of the teapot, the Simoneffect was produced.
Evidence has favored location coding over motoric factors inseveral earlier compatibility studies. Reeve and Proctor (1984) hadparticipants respond to a target stimulus, which could occur in anyof four locations on a row, by pressing the key in the correspond-ing location. Regardless of whether hand placement was adjacentor overlapped (index and middle fingers from the two handsalternated on the response keys), precuing the two left or two rightlocations was most beneficial. Miller (1982) found that in four-choice tasks with two-dimensional symbolic stimuli that have nospatial-location attribute, RT was shorter when the more salientstimulus feature distinguished responses on the left and righthands. He attributed this benefit to a feature of the motor systemthat allows more efficient preparation when the two responses“cued” by the salient dimension are on the same hand than whenthey are not. However, Proctor and Reeve (1985) showed that theobtained compatibility effects are similar to those shown by spatiallocation stimuli and that the salient feature benefit could be ob-tained even for between-hand responses. As a final example,Proctor, Van Zandt, Lu, & Weeks (1993) replicated a “destination”compatibility effect (a spatial compatibility effect based on theapparent destination of an object) that Michaels (1988) had ob-tained with “catching” movements of a joystick, showing that the
747CORRESPONDENCE EFFECTS FOR OBJECTS WITH PROTRUSION
results could be obtained with keypresses and regardless ofwhether the stimulus appeared to move toward or away from theparticipant. Even when destinations were designated by stationaryarrow stimuli, a similar compatibility effect was obtained, suggest-ing that that the destination compatibility effect is a function ofrelative location coding rather than catching affordance.
For the present study, in addition to the mean RT data, theresults obtained for the RT distributions are more consistent withrelative location coding than grasping affordance. The object prop-erty effect and the absence of within-hand Simon effect were notevident in the RT distribution data. All experiments showed theincreasing pattern regardless of whether the judgment or the re-sponse was action-related. Wiegand and Wascher (2005) providedevidence that increasing functions are not restricted to object-based Simon effects but also occur for location-based Simoneffects when the stimulus-location and response-location dimen-sions are both vertical. In comparison to the decreasing functionsoften found for left-right versions of the location Simon effect,they have argued that the “increasing-effect function is associatedwith the code interference process of more cognitive spatial codes”(Wiegand & Wascher, 2007, p. 402). Thus, an increasing distri-bution function is consistent with a spatial coding account.
With the teapot stimuli in Experiments 1–3 and the frying panstimuli in Cho and Proctor’s (2010) study we repeatedly obtainedobject-based Simon effects with color judgments that were similarin size and other respects to the effects obtained with orientationjudgments. Because Tipper et al. (2006) reported no Simon effectfor door-handle orientation with color judgments, we comparedour method to theirs and decided that a likely critical factor wasthat the handles were centered in their study but not in ours. InExperiment 4, therefore, we demonstrated a large Simon effect forcolor judgments with the door-handle stimuli when the base wascentered but not when the handle was centered. The finding of anobject-based Simon effect with the door-handle stimuli when thebase is centered indicates that our demonstrations of such effectswith color judgments for teapots and frying pans are not restrictedto the specific conditions of our experiments including use ofobject depictions rather than photographs and use of only fourstimulus alternatives instead of 16 or more. The absence of such aneffect when the handle is centered likely is a consequence of theonly change in handle position from trial to trial being horizontal(passive state) or diagonal (active state) orientation, rather than leftor right.
Contrary to our explanation of the results of our experimentsusing a location coding account, it might be argued that the relativelack of evidence for a grasping affordance effect is due to thehandle of the teapot being too small in relation to the pot itself forgenerating a grasping affordance effect because the size of thegraspable part is an object property considered critical for grasp-ing. However, Cho and Proctor (2010) found no evidence ofgrasping affordance with the longer handle of the frying pan (5.5cm), which is more than twice the length of the teapot handle (2cm). Moreover, in comparison to the present experiments, thephysical size of the handle corresponded with the size of the Simoneffect. The overall Simon effect for intact frying pans in Cho andProctor’s experiments was 43 ms, which was 2.69 times largerthan the Simon effect in the present handle-only experiment (16ms). This ratio approximately corresponds with the ratio of the
physical size of the two handles (5.5 cm/2 cm � 2.75), againimplicating relative location of a spatially distinct component.
Even though our results generally support the relative locationcoding account, there are several questions and issues that remain.First, the use of a single symbolic figure of a teapot rather than ofpictures of multiple real teapots (or other objects) might preventactivation of a grasping affordance. However, the door-handlestimuli of Experiment 4 were based on photographs of objects, andthere were 16 different variations as opposed to only four stimuliin the teapot experiments. Yet, the Simon effect was obtained forcolor judgments in that experiment when the bases of the handleswere centered. Moreover, there is no a priori reason to think thatmultiple objects must be depicted or that the depictions must bephotographs of actual objects. In fact, the spout-only teapot stimuliin Experiment 2 generated a Simon effect of similar size to themore object-like door-handle stimuli in the base-centered condi-tion of Experiment 4, which were based originally on photographs.
Second, although Tucker and Ellis (1998) used keypress re-sponses to test their affordance account, as did the studies ofTipper et al. (2006) and Loach et al. (2008), left and right graspresponses may be more likely to show an influence of graspingaffordance because those responses are similar to one’s that wouldbe made when grabbing an object. However, prior results haveshown that left and right keypresses yield similar results to forwardmovements of left and right joysticks intended to be more similarto catching affordances (Proctor et al., 1993). Several recent stud-ies of object-based Simon effects for small and large objects usingprecision and power grasp devices (e.g., Derbyshire et al., 2006)have suggested a possible contribution of grasping affordance.However, correspondence of grasp types to object size may rep-resent only another instance of dimensional overlap (Kornblum etal., 1990) and not affordance effects of the type usually implied.The role of relative spatial relations and grasping affordances inobject-based Simon effects obtained with grasp responses shouldbe a focus of future research.
Conclusion
The reported experiments demonstrate that an object with leftand right protrusions (a teapot) produces an object-based Simoneffect that reflects the contributions of the opposing locationcomponents. The more salient spout determined the direction ofSimon effect when opposed by the handle and the largest Simoneffect when the handle was removed, although the handle aloneyielded a Simon effect when the spout was removed. These Simoneffects were evident for both upright-inverted orientation judg-ments and color judgments, as well as for both within- andbetween-hand responses, indicating that they are not restricted tojudgments related to grasping. Along with the results reported byCho and Proctor (2010) for frying pan stimuli, the findings indicatethat relative location coding is a major contributor to object-basedSimon effects when the objects have distinct lateral appendages.
References
Anzola, G. P., Bertoloni, G. G., Buchtel, H. A., & Rizzolatti, G. G. (1977).Spatial compatibility and anatomical factors in simple and choice reac-tion time. Neuropsychologia, 15, 295–302.
Bosbach, S., Prinz, W., & Kerzel, D. (2005). Is direction position?
748 CHO AND PROCTOR
Position- and direction-based correspondence effects in tasks with mov-ing stimuli. Quarterly Journal of Experimental Psychology, 58A, 467–506.
Bub, D. N., & Masson, M. E. J. (2010). Grasping beer mugs: On thedynamics of alignment effects induced by handled objects. Journal ofExperimental Psychology: Human Perception and Performance, 36,341–358.
Buckolz, E., O’Donnell, C., & McAuliffe, J. (1996). The Simon effect:Evidence of a response processing “functional locus.” Human MovementScience, 15, 543–564.
Cho, D. (T.), & Proctor, R. W. (2010). The object-based Simon effect:Grasping affordance or relative location of the graspable part? Journal ofExperimental Psychology: Human Perception and Performance, 36,853–861.
Craft, J. L., & Simon, J. R. (1970). Processing symbolic information froma visual display: Interference from an irrelevant directional cue. Journalof Experimental Psychology, 83, 415–420.
Derbyshire, N., Ellis, R., & Tucker, M. (2006). The potentiation of twocomponents of the reach-to-grasp action during object categorisation invisual memory. Acta Psychologica, 122, 74–98.
Gibson, J. J. (1979). The ecological approach to visual perception. Boston.Houghton Mifflin.
Heister, G., Ehrenstein, W. H., & Schroeder-Heister, P. (1987). Spatial S-Rcompatibility with unimanual two-finger choice reactions: Effects ofirrelevant stimulus location. Perception & Psychophysics, 42, 195–201.
Hommel, B. (1993a). The relationship between stimulus processing andresponse selection in the Simon task: Evidence for a temporal overlap.Psychological Research, 55, 280–290.
Hommel, B. (1993b). The role of attention for the Simon effect. Psycho-logical Research, 55, 208–222.
Hommel, B., & Lippa, Y. (1995). S-R compatibility effects due to context-dependent spatial stimulus coding. Psychology Bulletin & Review, 2,370–374.
Katz, A. N. (1981). Spatial compatibility effects with hemifield presenta-tion in a unimanual two-finger task. Canadian Journal of Psychology,35, 63–68.
Kornblum, S. (1965). Response competition and/or inhibition in two-choice reaction time. Psychonomic Science, 2, 55–56.
Kornblum, S., Hasbroucq, T., & Osman, A. (1990). Dimensional overlap:Cognitive basis for stimulus-response compatibility: A model and tax-onomy. Psychological Review, 97, 253–270.
Lamberts, K., Tavernier, G., & d’Ydewalle, G. (1992). Effects of multiplereference points in spatial stimulus-response compatibility. Acta Psy-chologica, 79, 115–130.
Loach, D., Frischen, A., Bruce, N., & Tsotsos, J. (2008). An attentionalmechanism for selecting appropriate actions afforded by graspable ob-jects. Psychological Science, 19, 1253–1257.
Michaels, C. F. (1988). S-R compatibility between response position anddestination of apparent motion: Evidence of the detection of affordances.Journal of Experimental Psychology: Human Perception and Perfor-mance, 14, 231–240.
Miller, J. (1982). Discrete versus continuous stage models of humaninformation processing: In search of partial output. Journal of Experi-mental Psychology: Human Perception and Performance, 8, 273–296.
Nicoletti, R., Anzola, G. P., Luppino, G., Rizzolatti, G., & Umilta, C.
(1982). Spatial compatibility effects on the same side of the bodymidline. Journal of Experimental Psychology: Human Perception &Performance, 8, 664–673.
Phillips, J. C., & Ward, R. (2002). S-R correspondence effects of irrelevantvisual affordance: Time course and specificity of response activation.Visual Cognition, 9, 540–548.
Proctor, R. W., & Reeve, T. G. (1985). Compatibility effects in theassignment of symbolic stimuli to discrete finger response. Journal ofExperimental Psychology: Human Perception and Performance, 11,623–639.
Proctor, R. W., & Reeve, T. G. (1986). Salient-feature coding operations inspatial precuing tasks. Journal of Experimental Psychology: HumanPerception and Performance, 12, 277–285.
Proctor, R. W., & Reeve, T. G. (1990). Research on stimulus-responsecompatibility: Toward a comprehensive account. In R. W. Proctor, &T. G. Reeve (Eds.), Stimulus-response compatibility: An integratedperspective (pp. 483–494). Amsterdam: North Holland.
Proctor, R. W., Van Zandt, T., Lu, C.-H., & Weeks, D. J. (1993). Stimulus-response compatibility for moving stimuli: Perception of affordances ordirectional coding? Journal of Experimental Psychology: Human Per-ception and Performance, 19, 81–91.
Proctor, R. W., & Vu, K.-P. L. (2006). Stimulus-response compatibilityprinciples: Data, theory and application. Boca Raton, FL: CRC Press.
Proctor, R. W., & Vu, K.-P. L. (2010). Stimulus-response compatibility formixed mappings and tasks with unique responses. Quarterly Journal ofExperimental Psychology, 63, 320–340.
Reeve, T. G., & Proctor, R. W. (1984). On the advance preparation ofdiscrete finger responses. Journal of Experimental Psychology: HumanPerception and Performance, 10, 541–553.
Reeve, T. G., & Proctor, R. W. (1988). Determinants of two-choicereaction-time patterns for same-hand and different-hand finger pairings.Journal of Motor Behavior, 20, 317–340.
Rubichi, S., Vu, K-P. L., Nicoletti, R., & Proctor, R. W. (2006). Spatialcoding in two dimensions. Psychonomic Bulletin & Review, 13, 201–216.
Tipper, S. P., Paul, M. A., & Hayes, A. E. (2006). Vision for action: Theeffects of object property discrimination and action state on affordancecompatibility effects. Psychonomic Bulletin & Review, 13(3), 493–498.
Tucker, M., & Ellis, R. (1998). On the relations between seen objects andcomponents of potential actions. Journal of Experimental Psychology:Human Perception and Performance, 24, 830–846.
Tucker, M., & Ellis, R. (2001). The potentiation of grasp types duringvisual object categorization. Visual Cognition, 8, 769–800.
Wallace, R. J. (1971). S-R compatibility and the idea of response codes.Journal of Experimental Psychology, 88, 354–360.
Wiegand, K., & Wascher, E. (2005). Dynamic aspects of stimulus-responsecorrespondence: Evidence for two mechanisms involved in the SimonEffect. Journal of Experimental Psychology: Human Perception andPerformance, 31, 453–464.
Wiegand, K., & Wascher, E. (2007). Response coding in the Simon task.Psychological Research, 71, 401–410.
Received August 27, 2009Revision received August 19, 2010
Accepted September 2, 2010 �
749CORRESPONDENCE EFFECTS FOR OBJECTS WITH PROTRUSION
The Object-Based Simon Effect:Grasping Affordance or Relative Location of the Graspable Part?
Dongbin (Tobin) Cho and Robert W. ProctorPurdue University
Reaction time is often shorter when the irrelevant graspable handle of an object corresponds with thelocation of a keypress response to the relevant attribute than when it does not. This object-based Simoneffect has been attributed to an affordance for grasping the handle with the hand to the same side. Becausea grasping affordance should differentially affect keypress responses only when they are made withdifferent hands, we conducted three experiments that measured the object-based Simon effect for fryingpan stimuli using between- and within-hand response sets. When the relevant stimulus dimension wascolor, neither the object-based Simon effect nor the location-based Simon effect varied across responsesets. When upright-inverted orientation judgments were made for the frying pan and for nongraspablestimuli derived from it, there again was no significant difference in size of the between- and within-handSimon effects for any of the stimuli. The results provide evidence that the Simon effect for graspablefrying pan stimuli is because of relative location of the handle and not to a grasping affordance.
Keywords: affordance, grasping, Simon task, stimulus-response compatibility, within-hand responses
The spatial stimulus-response compatibility (SRC) paradigm re-quires participants to make spatial responses (e.g., left and rightkeypresses) based on the location of a stimulus (e.g., left or right). Thespatial SRC effect is that responses are faster and more accurate whenthe stimulus and response locations correspond than when they do not(Alluisi & Warm, 1990). Spatial compatibility effects are a function ofthe relative locations of the alternative stimuli and responses (Nico-letti, Anzola, Luppino, Rizzolatti, & Umilta, 1982). Consequently,they are typically attributed to response selection being faster whenthe spatial codes for the stimuli and responses correspond than whenthey do not (Proctor & Reeve, 1990).
Spatial compatibility effects also occur when responding to arelevant nonspatial feature such as stimulus color: Response time(RT) is shorter when the irrelevant stimulus location corresponds withthe correct response location than when it does not, a phenomenoncalled the Simon effect (Simon, 1990). This phenomenon is typicallyattributed to automatic activation of the spatial response code corre-sponding to the spatial stimulus code because of overlap between thespatial dimensions of the stimulus and response sets (Kornblum,Hasbroucq, & Osman, 1990). This automatic activation producesfaster and more accurate responding when the activated response codecorresponds with the spatial location of the response signaled by therelevant stimulus dimension than when it does not. Spatial codingaccounts of this type can explain a range of findings concerning theSimon effect (e.g., see Lu & Proctor, 1995).
In contrast to the spatial coding view, Michaels (1988, 1993)argued that many SRC effects are based on the detection of
affordances, or possibilities for action (Gibson, 1979), in environ-ments containing relatively rich visual information. According toaffordance accounts, people’s everyday actions are controlled bydirect, meaningful relations between objects in the visual environ-ment and the action systems of the observer. Tucker and Ellis(1998) proposed the hypothesis that “the actions [an] object af-fords are automatically potentiated” (p. 833) and explored thishypothesis for graspable objects in a variation of the Simon task.Their basic idea was that rather than object orientation automati-cally generating a left-right spatial code by virtue of the object’svisual properties, the affordance for grasping by the left or righthand is what gives rise to the left-right response distinction.According to their hypothesis, the left or right code generated bya graspable object consists of the activation of motor patternslinked with a specific action of the hand that is compatible with thelocation of the handle of the object.
To support their affordance hypothesis, Tucker and Ellis (1998)provided evidence that the left or right location of the graspablepart of an object relative to the main part of the object produces aSimon-type effect for left-right push-button responses. In theirExperiment 1, participants made one response with the left handand the other with the right hand depending on whether the objectwas in an upright or inverted orientation. Responses were 12 msfaster and 3% more accurate when the graspable part of the objectcorresponded with the response than when it did not, though thelocation of that part was irrelevant to the orientation decision. Thisobject-based Simon effect is consistent with Tucker and Ellis’saffordance hypothesis. However, it could be a consequence ofrelative location coding instead because the graspable part of theobject was located to the left or right of the object’s body (e.g., seeAnderson, Yamagishi, & Karavia, 2002).
Consequently, Tucker and Ellis (1998) conducted a criticallyimportant Experiment 2 in which the Simon effect was predicted tobe absent by their affordance hypothesis but to occur by a location
Dongbin (Tobin) Cho and Robert W. Proctor, Department of Psycho-logical Sciences, Purdue University.
Correspondence concerning this article should be addressed to Dongbin(Tobin) Cho, Department of Psychological Sciences, Purdue University, 703Third Street, W. Lafayette, IN 47907-2081. E-mail: [email protected]
coding explanation. That experiment was the same as their Exper-iment 1 except that participants responded with the index andmiddle fingers of the right hand. Because both responses wereconsistent with a right-hand grasp, they did not differ with regardto a left or right grasping affordance, and thus the affordanceaccount predicts no Simon effect. In contrast, because the within-hand responses can still be coded as left or right relative to eachother, the location coding account predicts a Simon effect to occur.Consistent with the prediction of the grasping affordance account,the graspable objects yielded only a nonsignificant 3-ms object-based Simon effect. This lack of significant within-hand Simoneffect in this instance is in contrast to results for the location-basedSimon effect, for which the Simon effect is at least as large forwithin-hand keypresses as for between-hand keypresses (Buckolz,O’Donnell, & McAuliffe, 1996; Heister, Ehrenstein, & Schroeder-Heister, 1987; Katz, 1981).
Although the absence of a within-hand object-based Simoneffect is more unique than the presence of a between-hand effect,and more probative, this outcome has not been highlighted muchin subsequent studies. An exception is that of Phillips and Ward(2002), who noted the importance of Tucker and Ellis’s (1998)Experiment 2 for testing the relative contributions of spatial codingand affordances for grasping, concluding that the results “areprobably not conclusive” (p. 548). However, they then conductedtests that did not include a within-hand condition. Close scrutiny ofTucker and Ellis’s Experiment 2 is needed, however, because theirfinding of a nonsignificant object-based Simon effect for within-hand responses was equivocal. That experiment was conductedseparately from their experiment using between-hand responses,and, although the mean RT data showed no significant within-handSimon effect, the median RT data did. Moreover, their study didnot include comparison under similar circumstances of the within-and between-hand Simon effects for spatial stimulus location.
The main goal of the present study, therefore, was to measurethe within- and between-hand Simon effects for a graspable objectand to compare them to measures obtained in the standardlocation-based Simon task and with modified versions of theobject that were not graspable. If the object-based Simon effect isbecause of a grasping affordance, then that effect should be elim-inated when the two responses are within the same hand but thelocation-based Simon effect and those obtained with ungraspableobjects should not be. In contrast, if the object-based Simon effectis because of activation of a left-right spatial code, then both thateffect and the other Simon effects should be no smaller withinhands than between hands.
In addition to examining the overall mean Simon effects in thevarious conditions, we conducted analyses of changes in the Simoneffect across the RT distributions. The Simon effect for stimuluslocation typically peaks in size at the fast end of the distribution anddecreases as RT increases, a finding that is usually attributed to rapidactivation of the corresponding response that then dissipates (e.g., DeJong, Liang, & Lauber, 1994). Several lines of evidence suggest that,in contrast, the object-based Simon effect with keypress responseswill increase in size across the RT distribution. Phillips and Ward(2002) used a procedure in which a frying pan appeared at variousintervals before onset of a superimposed imperative stimulus thatrequired a left or right keypress. The object-based Simon effect wasnot evident when the frying pan and imperative stimulus onset simul-taneously, but the effect appeared and increased as the interval be-
tween frying-pan onset and stimulus increased up to 1,200 ms. Tuckerand Ellis (2001) and Derbyshire, Ellis, and Tucker (2006) performedRT distribution analyses for tasks that required power and precisiongrasp responses in the presence of objects for which one grasp or theother would be appropriate. The object-based correspondence effectfor grasp type increased about 40 ms from the first to fifth quintile ofthe RT distribution. Finally, Riggio et al. (2008; Experiment 1, onsetcondition) recently reported a distribution analysis for an experimentin which between-hand left-right keypresses were made to theupright-inverted orientations of kitchen objects with graspable han-dles, which were presented above and below a fixation cross. Theobject-based Simon effect increased from 7.5 ms at the two shortestRT quintiles to 17.5 ms at the two longest ones.
Thus, our expectation was that the object-based Simon effect inthe present study also would increase as overall RT increased,suggesting that the effect does not occur rapidly and automatically,but takes time to develop. If the increasing function is because ofactivation of a grasping affordance, then the increase in Simoneffect at longer RTs should not be evident for conditions in whicha grasping affordance could not be a factor.
In Experiment 1, participants performed either a location-basedor object-based Simon task, with the object being a frying pan withthe handle located to the left or right, using either the left and rightindex fingers or the index and middle fingers of the right hand. Wechose the frying pan stimulus for the object-based task because ithas only one component that extends from the object and thatcomponent, the graspable handle, is positioned to the left or rightof the pan itself. For both tasks, the relevant stimulus dimensionwas red or green color. Experiment 2 used the object-based Simontask, but with the relevant dimension being whether the pan was inan upright or inverted orientation, as in Tucker and Ellis’s (1998)study. In addition, two types of ungraspable stimuli derived fromthe frying pan were tested: For one, the handle tip was separatedfrom the pan; for the other, a dashed line corresponding to themidline of the handle was the only visible stimulus. Finally, inExperiment 3, the manipulation of within- versus between-handresponse mode for upright-inverted judgments of frying pans wasmade within subjects to provide a more powerful test of thepredicted difference within and between hands.
Experiment 1
Participants performed a standard Simon task for which stimuliappeared in left-right locations or an object-based Simon task inwhich a handle was located to the left or right side of a centeredfrying pan. In all conditions, left-right keypress responses weremade to the red or green color of the stimulus. Within each task,half of the participants used the index fingers of the left and righthands to respond, whereas half used the index and middle fingersof the right hand. On the basis of prior studies and spatial codingaccounts, we expected that the within-hand location Simon effectwould be at least as large as the between-hand Simon effect. Ofmost concern was whether that would also be the case for theobject-based Simon effect or whether that effect would be elimi-nated within hands, as Tucker and Ellis’s (1998) study suggests.
Method
Participants. There were 96 students enrolled in IntroductoryPsychology courses participated for credits toward a course re-
854 CHO AND PROCTOR
quirement. All participants reported having normal or corrected-normal vision and normal hearing, and they were naive as to thepurpose of the experiment. Sixty-four participants were testedinitially, being randomly assigned to one of four groups. Twogroups performed the standard Simon task with small coloredcircles presented in left and right locations, and two groups per-formed an object variant of the Simon task for which the stimuliwere centered green and red frying pans with handles located tothe left or right side. One group for each task used between-handresponses (the left and right index fingers) and the other groupused within-hand responses (the right index and middle fingers, asin Tucker and Ellis’s, 1998, Experiment 2). Because the circles forthe standard Simon task were smaller than the pan for the objectversion, an additional 32 participants, 16 for each response con-dition, performed the standard Simon task using large coloredcircles the same size as the pan.
Apparatus and materials. Stimuli were pictures and tonesgenerated by Micro Experimental Laboratory (MEL 2.01; Schnei-der, 1995), which controlled the experiment. For all conditions, theparticipant sat directly in front of the monitor, at a viewingdistance of approximately 60 cm. For the original Simon task withsmall stimuli, the stimuli were green and red circles (MEL colorcode �2 for green and �4 for red) of �1.2-cm diameter, presentedin left and right physical locations, with the center of each circle 8cm from the center of the screen. For the object-based Simon task,stimuli were red and green frying pans that were modified versions ofa photograph of a frying pan (Creative Chemistry, 2006). The bodywas 5.3-cm diameter and the handle 5.5-cm long (see Figure 1, topleft panel); the frying pan was displayed in the center of the screen,with the location of the handle being to the left or right. The handlewas located �45° from the center vertical line of the stimuli toeither side. For the Simon task with large stimuli, the circles werethe same size as the body of the frying pan, with the center of eachcircle being 5.5 cm from the center of the screen. In all conditions,responses were registered by presses of one of two adjacent keyson the bottom row of the computer keyboard (the B or N key) withthe appropriate finger.
Design and procedure. Each condition consisted of twoblocks of 16 practice trials followed by 176 test trials in whicheach object occurred equally often in each color and with a left orright attribute (stimulus location for the standard Simon task andhandle location for the object Simon task), with order randomizedfor each participant. Participants were told to make a left or rightresponse depending on whether the object was green or red color.They performed the first trial block with one color-response map-ping and the second block with the other, with the order of themappings counterbalanced across participants. The instructionsstated to respond as fast and accurately as possible, without mak-ing too many errors.
For the small circle condition, each trial began with onset of thefixation cross for 1,000 ms, as is often the case in studies of thestandard Simon task. The stimulus then appeared and remainedvisible until a response was made or for 1,500 ms, at which timethe trial was terminated if no response had been made. Theprocedure was similar for the frying pan and large circle condi-tions, except that no fixation cross was used, as in Tucker andEllis’s (1998) study. Participants were not given feedback onresponse latencies, but errors were immediately followed by a400-Hz tone for 500 ms from the computer. The screen was blankfor 500 ms following the response or error tone, and then the nexttrial began.
Results
Mean RT and PE. Mean RT for correct responses and per-centage error (PE) were computed for each participant (Table 1).Trials for which RT was �200 ms or the trial terminated after1,500 ms (�1.0% of all trials in this experiment and the others)were discarded from the subsequent analyses. A 2 (response mode:between hand, within hand) � 2 (correspondence: corresponding,noncorresponding) � 3 (condition: small circle; large circle; fryingpan) ANOVA was conducted for RT and PE, with repeated mea-sures on the first two factors. The mean RT and PE data are shownin Table 1.
Figure 1. Examples of the stimuli used in the experiments.
855OBJECT-BASED SIMON EFFECT
The overall mean RT and PE were 492 ms and 2.0%, respec-tively. RT did not differ significantly between the three stimulusconditions, F � 1, but error rate did, F(1, 90) � 7.11, MSe � 7.85,p � .001, �p
2 � .14: PE was highest in the small circle condition(M � 3.01%) and lowest in the frying pan condition (M � 1.15%),with that for the large circle condition being intermediate (M �1.97%). The other between-subject factor, response mode, had nomain effect for RT or PE, Fs � 1, though it tended to interact withcondition for PE, F(2, 90) � 3.07, p � .051, but not RT. Thisinteraction was due mainly to the error patterns shown for thesmall circle and frying pan conditions being opposite, F(1, 60) �6.12, MSes � 47.54, p � .05, �p
2 � .09, with PE tending to behigher between hands (M � 3.71%) than within hands (M �2.31%) for the small circles, but smaller between hands (M �0.63%) than within hands (M � 1.67%) for the frying pan. Thelarge circles showed little difference in PE between the two re-sponse modes (Ms � 2.92 and 3.0%).
Both RT and PE showed main effects of correspondence, Fs(1,90) � 162.71 and 21.70, MSes � 186 and 2.46, ps � .001, �p
2 �.64 and .19, yielding Simon effects of 28 ms and 1.05%. PEshowed no remaining effects, Fs � 1, but RT showed a significantinteraction of correspondence with condition, F(2, 90) � 3.16, p �.047, �p
2 � .07. The Simon effect for RT was larger in the fryingpan (28.5 ms) and large circle (28.5 ms) conditions than in thesmall circle condition (18 ms). The interaction of correspondencewith response mode was close to the .05 level, F(1, 90) � 3.56,p � .062, but the direction of the mean difference was not towarda larger Simon effect between hands (22 ms) than within hand (28ms). It is important to note that there was no three-way interactionof the variables for either RT or PE, Fs � 1, indicating that thepatterns of within- and between-hand Simon effects were similarfor the three stimulus conditions.
RT distribution analyses. For each condition, RTs for non-corresponding and corresponding trials were rank ordered anddivided into four bins. For each bin, a Simon effect was computedby subtracting mean RT for the corresponding trials from that forthe noncorresponding trials. An ANOVA with the four bins, threeconditions, and two response modes as factors was performed.
This ANOVA showed a main effect of bin and a two-way inter-action of bin with condition, F(3, 270) � 5.85 and F(6, 270) �11.91, MSe � 497.37, ps � .001, �p
2 � .06 and �p2 � .21.
To clarify this interaction, individual ANOVAs were conducted foreach condition. For the small and large circle conditions (see Figure 2,top left and center panels), the main effect of bin was significant, Fs(3,90) � 15.41 and 2.98, MSes � 641.91 and 523.15, ps � .001 and.040, �p
2 � .34 and �p2 � .09, respectively, and there was no interac-
tion with response mode, Fs � 1.0. As is typically found, the Simoneffect decreased as RT increased. For the frying pan condition (seeFigure 2, top right panel), the bin main effect was significant, F(3,90) � 10.12, MSe � 327.04, p � .001, �p
2 � .25, and there also wasno interaction with response mode, Fs � 1.22. However, in this case,the Simon effect was an increasing function of RT for both within-and between-hand responses.
Discussion
RT and PE showed main effects of correspondence, yielding asignificant Simon effect. The object-based Simon effect of thefrying pan condition was larger than the location-based Simoneffect with the small circle stimuli but similar to that with the largecircle stimuli. Most important, the Simon effect showed no ten-dency to be larger for the between-hands group than for thewithin-hand group, both for the circle stimuli and the frying pan.Thus, Experiment 1 showed no evidence that, when the relevantdimension is color, the object-based Simon effect is restricted tothe between-hand response condition or that it is even any largerfor that condition than for the within-hand condition. This result isin keeping with the hypothesis that the object-based Simon effectreflects coding of location of the handle relative to the pan ratherthan a grasping affordance.
As customarily obtained (e.g., De Jong et al., 1994), the Simoneffect for stimulus locations decreased across the RT distributions,and it did so similarly when responses were within hand as whenthey were between hand. Such functions indicate greater variabil-ity for the corresponding trials than for the noncorresponding ones(Zhang & Kornblum, 1997) and are often attributed to rapid
Table 1Mean Response Times (RT) and Percentage Errors (PE) as a Function of Correspondence and Response Mode (Between Hand vs.Within Hand), and the Simon Effect in Experiments 1–3
Experiment Condition Response mode
Corresponding Noncorresponding Simon effect
RT PE RT PE RT PE
1 Green/red small circle Between 505 3.4 519 4.1 14 0.7Within 486 2.0 508 2.7 22 0.7
Green/red large circle Between 462 1.4 485 2.5 23 1.1Within 476 1.1 510 2.9 34 1.8
Green/red frying pan Between 477 0.2 503 1.0 26 0.8Within 470 1.0 501 2.3 31 1.3
2 Upright/inverted frying pan Between 558 0.8 601 5.0 43 4.2Within 524 1.4 574 5.7 50 4.3
3 Upright/inverted frying pan (within-subject) Between 526 1.5 573 5.2 47 3.7Within 542 1.1 600 5.2 58 4.1
856 CHO AND PROCTOR
activation of the corresponding response, followed by dissipationof this activation (e.g., De Jong et al., 1994). The opposite findingof an increase in the Simon effect across the RT distribution for thefrying pan stimuli indicates greater variability for the noncorre-sponding trials than for the corresponding trials (Zhang & Korn-blum, 1997). This greater variability could be taken to suggest thatthe coding of relative position of the object handle does not occurimmediately and thus exerts more influence on those trials forwhich the participant is taking longer to respond. The pattern ofincreasing effects is in agreement with the distributional analysesfor power and precision grasp responses reported by Tucker andEllis (2001) and Derbyshire et al. (2006) and for keypress re-sponses reported by Riggio et al. (2008). It also provides evidencethat converges with the results of Phillips and Ward’s (2002)precuing study with keypress responses, which showed that theeffect of an object precue increased as the SOA between it and asuperimposed imperative stimulus lengthened.
Experiment 2
Experiment 1 differed from that of Tucker and Ellis (1998) inusing color as the relevant stimulus dimension for the object-basedSimon task, rather than up or down orientation of the graspableobject part. This difference in color versus orientation judgmentscould be the cause of the discrepant results. Symes, Ellis, andTucker (2005) found that the Simon effect for orientation of agraspable object part was evident when people had to classifyobjects by keypresses according to whether they were used in the
kitchen or garage (Experiment 1) but not when they had to classifythem as red or green color (Experiment 2). Similarly, Tipper, Paul,and Hayes (2006) found a Simon effect of door-handle directionwhen responses were based on the object’s shape but not whenthey were based on its color, leading Tipper et al. to conclude thataffordance-based motor activation did not occur for color discrim-inations.
If it is necessary to attend to an object’s shape for a graspingaffordance to be activated, then Tucker and Ellis’s (1998) findingof no within-hand Simon effect for object properties should bereplicated when the required judgments relate to an object’s shaperather than its color. Therefore, Experiment 2 was conductedsimilarly to the object-based condition of Experiment 1, but par-ticipants had to judge whether the pan handle was angled towardthe top or bottom (i.e., whether the pan orientation was upright orinverted), as in Tucker and Ellis’s study.
Two additional stimulus sets were tested that were similar to thefrying pans but without a grasping affordance. For one set, only thehandle tip and pan were visible. This disembodied handle-tipcondition is similar to one from Tipper et al.’s (2006) study, inwhich they found that the affordance-based effects they obtainedwere eliminated when the hinge/door attachment component wasremoved from door handles, making them not appear to be han-dles. For the remaining stimulus set, the stimuli were dashed linesat the same angles as the pan handles, presented without the pan.If a grasping affordance contributes to performance with the fryingpans when people are judging orientation, then the within-hand
Figure 2. Within- and between-hand Simon effects (in milliseconds) plotted as a function of the mean RT foreach quartile in the different task conditions of Experiments 1–3.
857OBJECT-BASED SIMON EFFECT
Simon effect should be smaller than the between-hand Simoneffect for the frying pan stimuli but not for the others.
Method
Ninety-six new students from the same subject pool as inExperiment 1 participated: Thirty-two received the frying panstimuli, 32 the disembodied handle-tip stimuli, and 32 the dashedline stimuli. For the frying pan group, the pans were gray (MELcolor code 7) with normal (handle oriented downward) or inverted(handle oriented upward) orientation (see Figure 1, upper rightquadrant). The body of the frying pan was centered on the screen,but the handle location and orientation (normal or inverted) variedrandomly. The handle was located �45° from the vertical line ofthe stimuli to the left or right, pointing upward (inverted) ordownward (normal). The disembodied handle-tip stimuli were alsogray frying pans but with disconnected handle tips (see Figure 1,lower left quadrant); these stimuli were derived from the intactfrying pan by deleting the main part of the handle. The dashed-linestimuli consisted of gray, dashed lines at locations correspondingto the center of the long axis of the pan handles (see Figure 1,lower right quadrant).
For all three groups, participants were evenly divided by randomassignment into between- and within-hand response groups. Par-ticipants were tested individually with the assigned stimulus setand response mode. They were instructed to make a left or rightkeypress depending on the orientation of the critical object featureand not its left-right location. For the frying pan stimuli, the initialinstruction referred to upright versus inverted, and the experi-menter noted that the handle was upward for the inverted pan anddownward for the upright pan. Because upright/inverted does notapply to the other two stimulus sets, the instructions for themindicated only upward versus downward orientation. For the dis-embodied handle-tip and dashed line stimulus sets, participantswere not told that the stimuli were derived frying pans. In otherrespects, the method was like that of Experiment 1.
Results
Mean RT and PE. Mean RT was 552 ms, and the overallerror rate was 2.9%; there was no difference between the tworesponse modes, Fs � 1.0. Both RT and PE showed effects ofcorrespondence, Fs (1, 90) � 394.5 and 163.3, MSes � 302.6 and4.12, ps � .001, �p
2 � .81 and �p2 � .65, yielding Simon effects of
50 ms and 3.75%. Neither measure showed an interaction betweencorrespondence and response mode, nor a three-way interaction ofthose variables with stimulus type, Fs � 1.0, except the two-wayinteraction for RT, F(2, 90) � 1.45, p � .23. With the intact fryingpan, the Simon effect was 43 ms and 4.2% for the between-handcondition and 50 ms and 4.3% for the within-hand condition; withthe disembodied handle-tip, the Simon effect was 50 ms and 3.5%for the between-hand condition and 56 ms and 3.3% for thewithin-hand condition; with the dashed-line condition, the Simoneffect was 49 ms and 4.1% for the between-hand condition and 53ms and 3.1% for the within-hand condition. No other terms weresignificant.
RT distribution analysis. An ANOVA of the RT distribu-tions similar to that of Experiment 1 showed a main effect of RTbin, F(3, 270) � 28.53, MSe � 551.95, p � .001, �p
2 � .24 and no
interaction with stimulus type, F � 1. All three conditions showedan increase in the Simon effect across RT bins [frying pan, F(3,90) � 7.04, MSe � 517.93, p � .001, �p
2 � .21(see Figure 2, middle row)], but no interaction with responsemode, Fs � 1.0.
Discussion
As expected, the disembodied handle-tip and dashed-line con-ditions, which offer no grasping affordance, showed within-handSimon effects of similar size to the between-hand Simon effects.More important, the intact frying pan condition also showed awithin-hand Simon effect that was at least as large as the between-hand Simon effect.1 In addition, all of the RT distribution func-tions showed the Simon effect to increase as RT increased, as wasthe case for the frying pan stimuli in Experiment 1 when colorjudgments were made. Thus, the increasing function is not a signof a grasping affordance, either. The results of Experiment 2 implythat a grasping affordance, which should have been available onlyfor the frying pans, did not contribute to performance, even whenthe judgments involved the object’s orientation.
Comparison of the intact frying pan conditions for Experiments1 and 2 was performed to determine whether the size of theobject-based Simon effect differed for color and upright-invertedjudgments. This comparison showed that the Simon effect wassmaller for both RT and PE for the color identification task ofExperiment 1 than for the upright-inverted task of Experiment 2,Fs(1, 60) � 10.47 and 23.57, p � .002 and p � .001, �p
2 � .15 and.28, and this difference in size did not interact with response mode,Fs � 1.0. Thus, although we obtained an object-based Simoneffect when the task required color judgments, the effect wassmaller than when the task required attending to the form of theobject, as in the studies of Symes et al. (2005) and Tipper et al.(2006). Why the frying pans showed a significant object-basedSimon effect in the present study but not in theirs is not entirelyclear. In the case of Symes et al., the lack of effect may have beenbecause of their use of many garage and kitchen objects, some ofwhich would not have as salient left-right features as those of thefrying pans. This possibility is not applicable, though, for thedoor-handle stimuli used by Tipper et al., which were similar inmany respects to the frying pans in having an elongated handle thatextended from a circular part that would attach the handle to adoor. Tipper et al. stated that their door-handle stimuli werepresented centrally on the screen, which would mean that thecircular attachment part was to the left of center for a handlepointing to the right and to the right of center for a handle pointingto the left. If this were the case, the location of the attachment partwould have countered any effect because of door-handle direction.
1 An initial experiment with the orientation judgments for the frying panstimulus showed a significant within-hand object-based Simon effect of 28ms, but this effect was significantly smaller than the 47-ms effect obtainedbetween hands. However, in all other experiments we have conducted, thetendency has been toward a slightly larger Simon effect within- thanbetween-hands, and the difference in effect sizes when that experiment iscombined with the frying pan conditions of Experiments 2 and 3 of thepresent article is 1 ms.
858 CHO AND PROCTOR
In contrast, in our experiments, the body of the pan was alwayscentered at fixation.
Experiment 3
Although the frying pan stimuli showed no evidence in Exper-iment 2 of a smaller Simon effect within- than between-hands, onemight argue that a small effect was not detected because responsemode was a between-subjects variable. Consequently, we con-ducted an additional experiment with the frying pan stimuli usinga within-subject design to increase statistical power.
Method
Thirty-two new participants from the same pool as in the pre-vious experiments were tested. The procedure was identical to thatof Experiment 1 and of the intact frying pan condition of Exper-iment 2, with the major change being that each participant per-formed one block of trials using the index fingers on each hand andone block using the index and middle fingers on the right hand.Order of these conditions was counterbalanced across participants,as was the mapping of upright-inverted orientations to responses,which remained the same for both trial blocks. The total number oftrials remained the same as in the earlier experiments, such thateach participant received 176 trials in each of the two response-condition blocks. Before each block, 16 practice trials were ad-ministered.
Results and Discussion
The mean RT was 560 ms, and the overall error rate was 3.3%.Within-hand responses showed longer RT than between-hand re-sponses (571 and 550 ms, respectively), F(1, 31) � 4.57, MSe �306.4, p � .04, �p
2 � .14, but error rate did not differ, F � 1.0.Both RT and PE showed effects of correspondence, Fs(1, 31) �141.7 and 67.5, MSes � 306.4 and 7.11, ps � .001, �p
2 � .82 and.69, yielding Simon effects of 52 ms and 3.9%. Most important, theinteraction of the two variables was nonsignificant for both RT,F(1, 31) � 2.54, p � .12, and PE, F � 1.0. Note that the tendencyin the RT data was for the Simon effect to be larger within hand(58 ms) than between hands (47 ms). This tendency was notbecause of having performed with one of the response conditionsbefore the other, as comparison for the first condition received byeach participant showed mean values of 62 and 49 ms, respec-tively.
RT distribution analysis. This analysis, performed similarlyto those for the previous experiments, showed a main effect of RTbin, F(3, 93) � 9.15, MSe � 1,335, p � .001, �p
2 � .23, but nointeraction with response mode, Fs � 1.2 (see Figure 2, bottomrow). Thus, the results are similar to those of Experiments 1–3 inshowing an increase in the Simon effect for the frying pan stimuliat longer RTs.
General Discussion
The present study confirms that the Simon effect occurs not onlywhen the irrelevant dimension is stimulus location but also whenit is the left or right relative location of a salient object property,in this case the handle of a frying pan. The Simon effect was
evident for all stimuli regardless of whether responses were madewith fingers on the same or different hands.
In most studies of the Simon effect, the left and right indexfingers are used for responding, as in the between-hand conditionsof the present experiments. The size of the between-hand Simoneffect ranged from 14 to 50 ms, being smallest for the small circlestimuli presented in left and right locations in Experiment 1 andlargest for orientation judgments with all of the variations of fryingpan stimuli in Experiments 2–4. These results suggest that relativelocation of the handle or modified handle contributes to the Simoneffect. This component could possibly be a grasping affordance forthe intact frying pan, but that is unlikely the case for the fryingpans in which the handle tip is detached or the only stimulus is adashed line angled upward or downward.
Of more interest are the Simon effects obtained when theresponses were the index and middle fingers of the right hand. Inagreement with previous studies, for Experiment 1 the Simoneffect for red and green circles in left and right locations was ofsimilar size for within-hand and between-hand responses. Thisresult pattern also held for the intact frying pan when the relevantstimulus dimension was color. That the Simon effect for the intactfrying pan was no smaller within- than between-hands when therelevant dimension was color suggests that color can be processedseparately from form, as in Tipper et al.’s (2006) and Symes etal.’s (2005) studies with door handles and a variety of garage andkitchen objects. However, when the relevant dimension was up-right/inverted orientation in Experiments 2 and 3, the Simon effectstill was no smaller within than between hands. This similarity ofwithin- and between-hand Simon effects also occurred when thetip of the handle was detached from the frying pan or the stimuluswas a dashed line corresponding in location and orientation to thehandle. On the whole, these results provide evidence that, counterto the grasping affordance account, the Simon effect for fryingpans with graspable handles is no smaller within- than between-hands and no different from the effect obtained with modifiedfrying pans that would not afford grasping.
In Tucker and Ellis’s (1998) study, there was no significantobject-based Simon effect when the two responses were on thesame hand. In contrast, our Experiments 2–3 showed large within-hand Simon effects. As noted in the Introduction, although Tuckerand Ellis’s mean RT data did not show a significant within-handSimon effect, a small significant effect was evident when medianRT was analyzed instead. A possible reason why the within-handeffect was larger in our study than in theirs is that Tucker and Ellisused a variety of graspable objects, some of which did not have ahandle that was as salient relative to the rest of the object as for thefrying pan (e.g., a teapot, for which the hand is curved and a spoutprotrudes from the opposite side).
The results obtained for the RT distributions are also difficult toreconcile with the grasping affordance hypothesis. The object-based Simon effect increased as RT increased when frying panswere judged as to color (Experiment 1) and orientation (Experi-ments 2 and 3), but it did so equally for within-hand responses asfor between-hand responses. Moreover, the Simon effect for thedetached handle-tip stimuli and dashed line stimuli in Experiment2 also showed the Simon effect to be an increasing function of RT.Thus, the pattern of increasing Simon effect as overall RT in-creases is not a signature of a grasping affordance.
859OBJECT-BASED SIMON EFFECT
If the increasing function is not because of a grasping affor-dance, what is it because of? Increasing functions have beenreported for location-based Simon tasks in which the irrelevantstimulus dimension is vertical location and the responses are alsoarrayed vertically (Weigand & Wascher, 2005). These authorsproposed that the Simon effect for vertical and horizontal dimen-sions have two different causes: The horizontal Simon effect,which decreases across the RT distribution, is because of rapidactivation of the corresponding response by automatic visuomotorinformation transmission; the vertical Simon effect is because ofcode interference in intentional S-R translation. Applied to thepresent findings, such an account would attribute the Simon effectfor the frying pan stimuli and their variations to the slower,intentional translation process. For all except the dashed-line stim-uli, this could reflect coding the location of the handle or handle-tip relative to the pan itself. That the dashed-line stimuli show asimilar result suggests that having to judge the relevant dimensionalong the vertical axis may play a role as well.
The results of the study as a whole indicate clearly that theobject-based Simon effect obtained with keypress responses tofrying pan stimuli is a consequence of relative location coding andnot a grasping affordance. This conclusion implicating the impor-tance of relative location coding is similar to one reached byBosbach, Prinz, and Kerzel (2005) for effects of task-irrelevantmotion information on left-right manual responses. Bosbach et al.reported several experiments showing the motion effects to be aconsequence of participants coding shifts in position relative to thestarting position and not to the motion being processed as direc-tional information. The results of our studies and theirs are inagreement that relative location coding tends to predominate in avariety of situations in which stimuli have spatial properties. Adifference of our experiments from similar studies of object-basedSimon effects is use of only a single object with a very distinctleft-right feature. The larger Simon effects obtained in the presentstudy than in previous ones is probably because of this difference,but it seems likely that relative location coding is the main factorin the object-based Simon effect for keypress responses with otherstimuli as well.
Several recent studies of object-based Simon effects haveused precision and power grasp responses instead of left-rightkeypresses (e.g., Derbyshire et al., 2006). Those studies havefound Simon effects based on the match of the grasp type withthe size of the graspable object. The role of relative location inobject-based Simon effects with grasp responses is less obviousthan in those obtained with left-right keypresses. However, thisless obvious role does not necessarily mean that the grasp-typeeffects are not also due primarily or exclusively to spatialrelations between the stimuli and response rather than to spe-cific grasping affordances. Studies similar to the present oneand that of Bosbach et al. (2005) are necessary to distinguishthese alternatives.
References
Alluisi, E. A., & Warm, J. S. (1990). Things that go together: A review ofstimulus-response compatibility and related effects. In R. W. Proctor &T. G. Reeve (Eds.), Stimulus response compatibility: An integratedperspective (pp. 3–30). Amsterdam: North Holland.
Anderson, S. J., Yamagishi, N., & Karavia, V. (2002). Attentional pro-
cesses link perception and action. Proceedings of the Royal Society ofLondon, Series B, 269, 1225–1232.
Bosbach, S., Prinz, W., & Kerzel, D. (2005). Is direction position? Position-and direction-based correspondence effects in tasks with moving stimuli.Quarterly Journal of Experimental Psychology, 58A, 467–506.
Buckolz, E., O’Donnell, C., & McAuliffe, J. (1996). The Simon effect:Evidence of a response processing “functional locus.” Human MovementScience, 15, 543–564.
Creative Chemistry (2006, December). Research Intelligence, 30. Re-trieved from http://www.liv.ac.uk/researchintelligence/issue30/biocoatings.html
De Jong, R., Liang, C.-C., & Lauber, E. (1994). Conditional and uncon-ditional automaticity: A dual-process model of effects of spatialstimulus–response correspondence. Journal of Experimental Psychol-ogy: Human Perception and Performance, 20, 731–750.
Derbyshire, N., Ellis, R., & Tucker, M. (2006). The potentiation of twocomponents of the reach-to-grasp action during object categorisation invisual memory. Acta Psychologica, 122, 74–98.
Gibson, J. J. (1979). The ecological approach to visual perception. Boston:Houghton Mifflin.
Heister, G., Ehrenstein, W. H., & Schroeder-Heister, P. (1987). Spatial S-Rcompatibility with unimanual two-finger choice reactions: Effects ofirrelevant stimulus location. Perception & Psychophysics, 42, 195–201.
Katz, A. N. (1981). Spatial compatibility effects with hemifield presenta-tion in a unimanual two-finger task. Canadian Journal of Psychology,35, 63–68.
Kornblum, S., Hasbroucq, T., & Osman, A. (1990). Dimensional overlap:Cognitive basis for stimulus-response compatibility: A model and tax-onomy. Psychological Review, 97, 253–270.
Lu, C.-H., & Proctor, R. W. (1995). The influence of irrelevant locationinformation on performance: A review of the Simon and spatial Stroopeffects. Psychonomic Bulletin & Review, 2, 174–207.
Michaels, C. F. (1988). S-R compatibility between response position anddestination of apparent motion: Evidence of the detection of affordances.Journal of Experimental Psychology: Human Perception and Perfor-mance, 14, 231–240.
Michaels, C. F. (1993). Destination compatibility, affordances, and codingrules: A reply to Proctor, Van Zandt, Lu, and Weeks. Journal of Experi-mental Psychology: Human Perception and Performance, 19, 1121–1127.
Nicoletti, R., Anzola, G. P., Luppino, G., Rizzolatti, G., & Umilta, C.(1982). Spatial compatibility effects on the same side of the bodymidline. Journal of Experimental Psychology: Human Perception andPerformance, 8, 664–673.
Phillips, J. C., & Ward, R. (2002). S-R correspondence effects of irrelevantvisual affordance: Time course and specificity of response activation.Visual Cognition, 9, 540–548.
Proctor, R. W., & Reeve, T. G. (1990). Research on stimulus-responsecompatibility: Toward a comprehensive account. In R. W. Proctor &T. G. Reeve (Eds.), Stimulus-response compatibility: An integratedperspective (pp. 483–494). Amsterdam: North Holland.
Riggio, L., Iani, C., Gherri, E., Benatti, F., Rubichi, S., & Nicoletti, R.(2008). The role of attention in the occurrence of the affordance effect.Acta Psychologica, 127, 449–458.
Schneider, W. (1995). MEL professional users guide. Pittsburgh, PA:Psychology Software Tools.
Simon, J. R. (1990). The effects of an irrelevant directional cue on humaninformation processing. In R. W. Proctor & T. G. Reeve (Eds.),Stimulus–response compatibility. An integrated perspective (pp. 31–86).Amsterdam: Elsevier.
Symes, E., Ellis, R., & Tucker, M. (2005). Dissociating space-based andobject-based affordances. Visual Cognition, 12, 1337–1361.
Tipper, S. P., Paul, M. A., & Hayes, A. E. (2006). Vision for action: Theeffects of object property discrimination and action state on affordancecompatibility effects. Psychonomic Bulletin & Review, 493–498.
860 CHO AND PROCTOR
Tucker, M., & Ellis, R. (1998). On the relations between seen objects andcomponents of potential actions. Journal of Experimental Psychology:Human Perception and Performance, 24, 830–846.
Tucker, M., & Ellis, R. (2001). The potentiation of grasp types duringvisual object categorization. Visual Cognition, 8, 769–800.
Weigand, K., & Wascher, E. (2005). Dynamic aspects of stimulus-responsecorrespondence: Evidence for two mechanisms involved in the SimonEffect. Journal of Experimental Psychology: Human Perception andPerformance, 31, 453–464.
Zhang, J., & Kornblum, S. (1997). Distributional analysis and De Jong andLiang and Lauber’s (1994). dual-process model of the Simon effect.Journal of Experimental Psychology: Human Perception and Perfor-mance, 23, 1543–1551.
Received April 29, 2008Revision received April 27, 2009
Accepted May 5, 2009 �
Members of Underrepresented Groups:Reviewers for Journal Manuscripts Wanted
If you are interested in reviewing manuscripts for APA journals, the APA Publications andCommunications Board would like to invite your participation. Manuscript reviewers are vital to thepublications process. As a reviewer, you would gain valuable experience in publishing. The P&CBoard is particularly interested in encouraging members of underrepresented groups to participatemore in this process.
If you are interested in reviewing manuscripts, please write APA Journals at [email protected] note the following important points:
• To be selected as a reviewer, you must have published articles in peer-reviewed journals. Theexperience of publishing provides a reviewer with the basis for preparing a thorough, objectivereview.
• To be selected, it is critical to be a regular reader of the five to six empirical journals that are mostcentral to the area or journal for which you would like to review. Current knowledge of recentlypublished research provides a reviewer with the knowledge base to evaluate a new submissionwithin the context of existing research.
• To select the appropriate reviewers for each manuscript, the editor needs detailed information.Please include with your letter your vita. In the letter, please identify which APA journal(s) youare interested in, and describe your area of expertise. Be as specific as possible. For example,“social psychology” is not sufficient—you would need to specify “social cognition” or “attitudechange” as well.
• Reviewing a manuscript takes time (1–4 hours per manuscript reviewed). If you are selected toreview a manuscript, be prepared to invest the necessary time to evaluate the manuscriptthoroughly.
861OBJECT-BASED SIMON EFFECT
Computers in Industry 61 (2010) 311–317
Influence of the Privacy Bird1 user agent on user trust of different web sites
Kim-Phuong L. Vu a, Vanessa Chambers a, Beth Creekmur a, Dongbin Cho b, Robert W. Proctor b,*a California State University Long Beach, United Statesb Purdue University, United States
A R T I C L E I N F O
Article history:
Available online 8 January 2010
Keywords:
E-commerce
Privacy practices
User agents
User trust
Web privacy
A B S T R A C T
With increasing growth of Internet commerce, online fraud accounts for as much as 20% of identity theft
cases. The present study evaluated Privacy Bird1, a computer program that warns users of privacy
preference violations by displaying a colored bird. Users rated their trust of, and willingness to give
financial information to, web sites in three categories (financial, retail, and social networking) before and
after using Privacy Bird. Privacy Bird improved participants’ privacy practices, increasing their trust in
(and willingness to provide financial information to) web sites that yielded green birds, reducing it for
sites that yielded red birds, and inducing further consideration of policies for sites that yielded yellow
birds. These results suggest that e-commerce sites should address the privacy concerns of users and
make salient the cues that inform users that their privacy is protected.
� 2009 Elsevier B.V. All rights reserved.
Contents lists available at ScienceDirect
Computers in Industry
journa l homepage: www.e lsevier .com/ locate /compind
1. Introduction
The success of an organization is often dependent on the trust ofits clients and customers. Consumer trust usually revolves aroundissues of securing information that consumers provide andmaintaining privacy protection. Privacy has the potential to beviolated any time that personal information is provided to anorganization. This potential for violation is amplified when thatinformation is sent over the World Wide Web, where it may bemaintained indefinitely in electronic forms. Once the personalinformation is provided, the individual user has no control overwho within the organization has access to the information, howthat information will be secured from attacks by unauthorizedparties, and whether it will be shared with other organizations.
Organizations that collect personally identifiable informationoften share the consumers’ personal information with affiliatesand third parties. Most consumers prefer that their personalinformation not be distributed in this manner [1], and anorganization would be wise to have a privacy policy that precludessharing of information. Regardless of the privacy policy adopted,though, many organizations do not provide sufficient securitymeasures to ensure that the personal information is protected. As aconsequence, breaches in security are common that result inpersonal information such as Social Security numbers, driver
* Corresponding author at: Department of Psychological Sciences, Purdue
University, 703 Third St., W. Lafayette, IN 47907, United States.
0166-3615/$ – see front matter � 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.compind.2009.12.001
license numbers, and credit card numbers being compromised. Forexample, Privacy Rights Clearinghouse (http://www.privacyrights.org/ar/ChronDataBreaches.htm#1) provides a chronology of databreaches from January 2005, to the present, which is updatedbiweekly. As of January 19, 2009, the chronology listed251,164,141 entries for the U.S. alone.
Two examples from this site are as follows. A listing datedAugust 5, 2008, indicates that a former employee of CountrywideFinancial Corporation, in Calabasas, California, was arrested andcharged with stealing and selling personal information, includingSocial Security numbers, over a 2-year period ending in July. Thenames of 2,000,000 people and their personal information weresold to other individuals in the mortgage industry for marketingpurposes. An August 12, 2008, listing states that Wells Fargo banknotified 5000 customers that hackers illegally accessed theirpersonal data, which included names, addresses, dates of birth,Social Security numbers, and driver’s license numbers. In the firstexample the security breach occurred from within the organiza-tion, and in the second it occurred from outside. Regardless, in bothcases, there was unauthorized access of personal data. Unautho-rized access can have negative consequences for the individualthrough such activities as identity theft and unauthorized chargesto a credit card, as well as for the organization and industry itrepresents (in these examples, financial institutions) by creatingconsumer distrust.
Web-based services are a growing industry due to the wideavailability of the Internet and the convenience of being able tomake online transactions. Because many web sites collectpersonally identifiable information of the type described above,either by necessity or choice, questions concerning how privacywill be handled and protected should be salient to both the
K.-P. Vu et al. / Computers in Industry 61 (2010) 311–317312
organizations that provide the services and the end users.Fortunately, many sites that collect personal information postprivacy policies that users can access [2], which describe how theorganization intends to use, store, and protect informationcollected through their site. This information is of value to usersbecause it allows them to make an informed choice as to whetherthe organization’s stated policy matches their own privacypreferences.
However, these privacy policies are typically written at a higheducational level [1,3,4] and are often formatted in small font andpositioned ‘‘below the fold’’ [4,5], making them difficult to locate andread. Moreover, even experienced computer and Internet users havedifficulty comprehending the privacy policies and are often unableto locate information within a policy that speaks to specific privacyconcerns [6]. Given that privacy policies are difficult for users tolocate and comprehend, it is not surprising that analyses of log filesshow that users seldom visit the privacy policy pages of web sites [4].This failure to inspect privacy policies is unfortunate because thepolicies are often the only means that a user has for discerning howan organization intends to use her or his personal, behavioral, andfinancial information after it is collected at their web site.
If users do not rely on a web site’s privacy policy to determinewhether to enter and submit personal information, on what basisdo they decide which web sites to use? Vu et al. [5] found that, fore-commerce web sites, users’ often relied on cheap prices,familiarity with the organization’s name, and past experiencewith the site to determine whether to make purchases. Forexpensive purchases, the company’s name was the most importantfactor. Unfortunately, it is not always the case that well-knowncompanies have better privacy practices then less well-knowncompanies. For inexpensive purchases, price of the item was themost important factor. Users will submit their personal and creditcard information to unfamiliar sites if the ‘‘price is right’’ due to thesmaller perceived risks associated with smaller purchases.However, users tend not to realize that regardless of the purchaseamount, the same information is being provided to the company.
In an attempt to empower users to protect their personalinformation, the platform for privacy preferences (P3P) wasdeveloped by the World Wide Web Consortium (W3C, seehttp://www.w3.org/P3P/). P3P is a method used to specify howpersonal information is collected and used by a web site in amachine-readable format. This format enables a user-agent tool tocheck a privacy policy for specific privacy concerns and report itsfindings to the end user. Ideally, the summary provided by the useragent will be in a format that users will be able to comprehendeasily. An effective user agent thus will mitigate the problem ofusers tending not to read privacy policies because of the policiesbeing long and difficult to understand.
There are browsers and user agents that can block cookies(packets of information sent to the web server) or inform a user ofpolicy concerns [7]. One freely available user agent is Privacy Bird1
(http://www.privacybird.org/), which assesses a web site’s privacypolicy for violations of specific privacy concerns. The user can setPrivacy Bird at one of three preset levels of security (low, medium,or high), or configure individual privacy specifications in severaldifferent categories. In a matter of seconds after a web page isloaded, Privacy Bird will return a colored bird in an unobtrusivelocation at the top of the browser, which conveys the results of itsassessment of the web site’s policy in relation to the privacycriteria set by the user. It employs traditional color-codingschemes: green for good (i.e., the policy satisfies the user’s privacypreferences), yellow for unknown (i.e., the site does not have amachine-readable policy posted), and red for dangerous (state-ments in the policy violate the user’s privacy preferences). Inaddition, a short policy summary for the web site is available tousers through the Privacy Bird menu.
Cranor et al. [8] showed that initial adopters of Privacy Birdengaged in better privacy practices than they had before, byreading privacy policies and taking steps to protect their personalinformation through continued use of Privacy Bird. Proctor et al.[1] found that Privacy Bird was able to check for 8 of the top 10privacy concerns identified by users in their study, making it auseful tool to aid users in deciding whether to provide informationto specific web sites. One question that remains, though, iswhether use of Privacy Bird can change users’ perceived trust of aweb site and their willingness to provide personal information todifferent sites. This question was investigated in the present study.
All research methods have their unique strengths and weak-nesses, which requires a multimethod approach to obtain acomplete understanding of most applied problems [9]. For thepresent study, we chose to obtain subjective ratings related toprivacy but in a laboratory setting that allowed us to manipulatecategory of web site, hold the specific sites constant, and control fororder effects. Even though such measures may deviate from thebehaviors observed in a naturalistic environment, they provideinsight into participants’ perceptions of privacy and the impact ofsoftware tools like Privacy Bird on their perceptions. A fullerpicture of the value of Privacy Bird and similar user agents willrequire that the findings of the present study be integrated withthose of other studies whose methods have different strengths andweaknesses.
Vu et al. [5] showed that participants do not use a web site’sprivacy policy to determine whether to make online transactions,but rely instead on familiarity with the site, the price of products,and name recognition. The present study evaluated whether use ofPrivacy Bird can positively influence users’ stated willingness toprovide personal information to sites that do or do not violate theirprivacy concerns and whether their trust levels change for sitesthat yield different bird colors. Three categories of web sites thatrequire the exchange of personal and financial information wereused in the study: financial/banking, retail, and social networking.These three categories allowed us to examine whether thewarnings provided by Privacy Bird influenced user onlinebehaviors when different perceived risks (e.g., financial vs.nuisance) were present. Participants configured Privacy Bird tomatch their privacy preferences and then evaluated individual websites with regard to whether (a) the site violated their privacypreferences, (b) they trusted the site, and (c) they would conducttransactions with the site.
We hypothesized that participants would be more likely totrust, and give their personal information to, sites that sellproducts and involve financial transactions than sites that providesocial services. This prediction of higher trust ratings for retail andfinancial sites is made because, even though they are more likely toinvolve financial risks than the social sites, users view personalinformation as more necessary to monetary transactions. Theamount of trust in the different categories of web sites washypothesized to be affected by the warning given by Privacy Bird.Trust ratings were predicted to increase if the site did not violatethe participant’s privacy preferences and decrease if it did. When ayellow bird indicates that whether the site violates privacypreferences is unknown, users should become more cautious, asindicated by less willingness to conduct transactions with the site.
2. Methods
2.1. Participants
A total of 20 undergraduate students were recruited, 10 fromCalifornia State University Long Beach [2 males and 8 females;mean (M) = 20 years of age; M = 14 years of education] and 10 fromPurdue University (3 males and 7 females; M = 23 years of age;
Capital One Comerica Reunion Tickle Circuit City PC Mall
Bank of America Key Corp Myspace The Student Center Fry’s Newegg
Peachtree RBC Royal Bank Classmates De.li.cious Radioshack Etronics
1 These data were only recorded for the CSULB subjects because the Purdue
subjects were told to check the privacy policy every time a yellow bird appeared and
just the privacy summary every time a red bird appeared. Institution did not
interact with any variable on willingness to give financial information and was
significant in only one interaction for the user trust ratings. Thus, this procedural
difference had little if any influence on the findings.
K.-P. Vu et al. / Computers in Industry 61 (2010) 311–317 313
M = 16 years of education). Participants reported being experi-enced with the Internet and computers (Ms = 1.5 and 1.1,respectively, on a scale for which 1 was very experienced and 4no experience), but not with privacy-policy issues (M = 2.5, with 1being very knowledgeable and 4 not knowledgeable) or withprograms designed to check privacy policies (M = 3.2). Participantswere paid $10 an hour for participation or received research creditstoward an Introductory Psychology course requirement.
2.2. Apparatus
This study was conducted on personal computers using InternetExplorer (version 6.0). Privacy Bird (version 1.3.0) was used tocheck and warn users about privacy-policy practices of the sitesthat were visited. Privacy Bird allows users to select from 12common privacy concerns grouped as (a) health and medicalinformation, (b) financial or purchase information, (c) personallyidentifiable information, and (d) non-personally identifiableinformation. Alternatively, users can select from low, medium,or high privacy presets, which specify increasingly larger sets ofitems from the 12 privacy concerns. Once concerns have beenspecified, Privacy Bird reviews machine-readable privacy policiesfor visited sites and conveys whether the user’s concerns may beviolated by displaying a green, yellow, or red colored bird.
2.3. Materials
Eighteen web sites representing a combination of bird color(green, yellow, or red when a medium preset level of privacyprotection was used), type of site (financial, retail, social), andrelative familiarity (unfamiliar, familiar) were visited (see Table1). Relative familiarity of web sites was determined prior to thestudy by referring to survey data [5], web site traffic ratings likeAlexa (www.alexa.com), and consensus of the researchers.However, because participants were able to set their own levelsof privacy protection with Privacy Bird and varied with regard towhether they indicated familiarity with particular sites, birdcolor and familiarity were recoded for the data analyses torepresent the actual bird color that a site yielded based on eachparticipant’s settings and the participant’s familiarity rating foreach site.
2.4. Procedure
Each participant was tested individually in a single sessionlasting between 1 and 1.5 h. Participants were given a pre-questionnaire that asked them whether they would join or performan online transaction with each site in a listing of financial, retailand social web sites. They were also asked to rate their familiarity,trust, and willingness to give financial information to each of thesesites on a scale of 1–7, with 1 being low and 7 being high.Afterwards, participants received a description of the Privacy Birdprogram and a demonstration of Privacy Bird (i.e., how the birdcolor changed depending on the privacy settings provided). Afterthe demonstration, participants were told that they would bevisiting several web sites in each of the three categories and that
they were to imagine they were interested in becoming a customerof each site, regardless of whether they were already a customer.Sites were visited one at a time, and for each site, participants wereallowed to examine the site for as long as they wanted. They wereasked to determine whether they would use the site for itsintended purpose. Participants were instructed to look at whateverinformation they deemed pertinent to their decision to use the site.Participants were also asked to determine whether the siteviolated their privacy preferences. After reaching a decision aboutuse of the current site, the participant verbally answered theexperimental questionnaire. This questionnaire consisted of thefollowing four questions about the site, for which the experimenterrecorded their answers:
1. Does the site violate your privacy preferences? Explain.2. Would you perform a transaction with the site?3. How much would you trust the site with your personal
information?4. How likely would you be to give your financial/credit card
information to the site?
The last two questions were answered on a scale of 1 (low) to 7 (high).All of the sites within a given category were viewed and judged
before moving on to the next category. Order of the threecategories was determined randomly for each participant, as wasthe ordering of the individual sites within each category.
3. Results
One retail site from the pre-experimental questionnaire (BestBuy) was not working on the day that the initial participants weretested and was replaced by another site (Fry’s), and one social site(De.li.cious) was unintentionally omitted from the pre-experi-mental questionnaire. For these reasons, those sites were excludedfrom the analysis of results that compare pre-experimental andexperimental questionnaires.
3.1. Privacy Bird settings and interpretation
For each of the three site categories, seven or eight participantsdecided to use the preset Privacy Bird settings, and the remainingparticipants chose the custom settings (see Fig. 1). For those whoused the presettings, nobody set Privacy Bird to low, a few set it tomedium (financial = 1; retail = 4; social = 2), and most set it to high(financial = 6; retail = 3; social = 6).
Participants were asked to indicate how they had determinedwhether a web site violated their privacy specifications:evaluating the color of the bird alone, the bird color plus thesummary provided by Privacy Bird, or the Privacy Policy itself1.As can be seen in Fig. 2, participants agreed that a green bird
Fig. 1. Categories selected for custom selection of Privacy Bird.
K.-P. Vu et al. / Computers in Industry 61 (2010) 311–317314
indicated no violation of privacy preferences and a red birdindicated violation of privacy preferences. For yellow birds, moreparticipants judged the sites as not violating their privacypreferences than violating them. For sites that yielded greenbirds, participants indicated that their decisions were most oftenbased on the bird color alone. For sites that yielded yellow birds,subjects were more cautious, looking at the full privacy policiesfor about half of the sites to determine whether their preferenceswere violated. For the sites that yielded red birds, the summaryprovided by Privacy Bird was often examined but not the fullprivacy policy.
Fig. 2. Percentage of total sites for which participants’ indicated no violation of
privacy preferences (top panel) or violation of privacy preferences (bottom panel)
as a function of bird color and whether this decision was based on the bird color
alone, the bird color plus summary, or the privacy policy.
3.2. Pre-experimental likelihood to use and familiarity judgments
Overall, 70% of the participants indicated that they would notuse or join the web sites in the study during the pre-questionnaire,with the proportion of sites for which the participants said ‘‘no’’being larger for the financial and social categories than for theretail category. This low usage rate may be due to the fact thatparticipants indicated that they were unfamiliar with 70% of thesites (see Table 2). The correlation with familiarity and willingnessto use the site was r = .55, p < .001. Thus, familiarity accounted for30% of the variance in willingness to use a site.
3.3. Willingness to give financial information
A 2 (time of questionnaire administration: pre vs. during theexperiment) � 3 (type: financial, retail, social) � 3 (bird color:green, yellow, red) � 2 (Institution: CSULB vs. Purdue) repeated-measures ANOVA was run on the ratings of willingness to givefinancial information to the web site. The results of the ANOVAare summarized in Table 3. Participants were more likely toprovide credit card information after (M = 3.25) interacting withthe site and seeing Privacy Bird’s rating than before (M = 2.65);however, even the ‘‘after’’ rating was low. Participants were alsomore likely to give their credit card information to sites thatyielded green (M = 3.39) and yellow (M = 3.11) birds than to sitesthat yielded red birds (M = 2.55). In addition, participants weremore likely to give their credit card information to financial(M = 3.35) and retail (M = 3.35) sites than to social networkingsites (M = 2.35).
Bird color influenced participants’ willingness to give finan-cial information from pre-experimental ratings to experimentalratings: willingness to provide financial information did notchange from pre-experimental to experimental ratings for redbird sites (M = 2.2 to M = 2.5) and green bird sites (M = 3.2 to 3.5),
Table 2Percentage of participants indicating willingness to join or use a site and familiarity
with a site in the pre-experimental questionnaire as a function of type of site.
Type of site Join/use Familiarity
Financial Yes 21% 18%
No 78% 81%
Retail Yes 40% 38%
No 60% 60%
Social Yes 29% 36%
No 71% 64%
Table 32 (time of questionnaire administration: pre vs. during the experiment)�3 (type:
Type of site�Bird color 4,72 3.96 1.20 .006Type of site� Institution 2,36 0.12 3.20 .883
Bird color� Institution 2,36 1.91 1.09 .162
Time�Type of site�Bird color 4,72 1.58 0.58 .189
Time�Type of site� Institution 2,36 0.26 1.08 .770
Time�Bird color� Institution 2,36 0.84 0.70 .441
Type of site�Bird color� Institution 4,72 2.70 1.20 .037Time�Type of site�bird color� Institution 4,72 0.67 0.58 0.62
K.-P. Vu et al. / Computers in Industry 61 (2010) 311–317 315
but yellow bird sites showed an increase in willingness ratings(M = 2.9 to 3.7). Bird color also influenced willingness to givecredit card information for the different types of sites: forfinancial sites, participants were more likely to give their creditcard information to sites with green (M = 3.9) or yellow (M = 3.9)birds than to sites with red birds (M = 2.3). For retail sites,participants were more likely to give their credit card informa-tion to sites with green birds (M = 3.7) than those with yellow(M = 3.2) or red birds (M = 3.1). For social sites, participants didnot want to give their financial information regardless of birdcolor (M = 2.5, 2.2, and 2.3, respectively, for green, yellow, andred birds).
The three-way interaction of questionnaire administration,type of web site, and bird color was significant. As illustrated inFig. 3, for sites with green and yellow birds, participants showedincreased willingness to give their credit card information frompre-experimental to experimental ratings, with the increase beinggreater for financial and retail sites than social sites. For sites withthe red bird, willingness for participants to give their credit carddecreased from pre-experimental to experimental ratings for allsite categories.
3.4. User trust
To determine if trust in a web site changed as a result ofPrivacy Bird influences, a 2 (time of questionnaire administra-
Fig. 3. Mean willingness to give financial information from pre-experimental vs.
experimental questionnaires by bird color and type of web site.
tion: pre vs. during the experiment) � 3 (type: financial, retail,social) � 3 (bird color: green, yellow, red) � 2 (institution:CSULB vs. Purdue) repeated-measures ANOVA was run on trustratings. The results of the ANOVA are summarized in Table 4.Participants trusted retail (M = 4.1) sites more than financial(M = 3.8) and social sites (M = 3.4), and sites with green (M = 4.1)or yellow (M = 4.0) birds more than sites with red (M = 3.1) birds.Bird color interacted with type of site, see Fig. 4. The decrease intrust for sites with green vs. red birds was greater for financial(Ms = 4.2 vs. 2.7 for green vs. red bird sites, respectively) than forretail (Ms = 4.4 vs. 3.7) and social (Ms = 3.9 vs. 2.9) sites. Thistwo-way interaction of bird color and type of site was modifiedby a three-way interaction of these variables with institution,with CSULB students showing a larger difference in trust ratingsbetween green and red bird sites only for financial sites, butPurdue students showing a difference for both financial andretail sites.
A difference in pre- and post-trust ratings was found as a resultof bird color, see Fig. 5. Whereas trust ratings for green bird sitesincreased from the pre-experimental questionnaire (M = 3.6) tothe experimental questionnaire (M = 4.7), and ratings for sites witha yellow bird remained relatively unchanged from pre- (M = 3.9)to experimental (M = 4.1) questionnaire, trust ratings for redbird sites decreased from the pre-questionnaire (M = 3.6) to theexperimental questionnaire (M = 2.6). No other effects weresignificant.
Fig. 4. Mean trust ratings by bird color for financial, retail, and social web sites.
Fig. 5. Mean trust ratings from pre-experimental vs. experimental questionnaires
by bird color.
K.-P. Vu et al. / Computers in Industry 61 (2010) 311–317316
4. Discussion
The dilemma in sharing personal information online isobvious. A certain amount of data exchange and storage maybe necessary for an organization to provide online services.Moreover, for frequently visited sites, stored information canlead to a more pleasant and efficient online experience for theuser (e.g., shopping at different e-commerce sites with saved userinformation). However, too much information collection resultsin undesirable feelings of privacy violation and calls intoquestion the motivation of the organization hosting the web site.
Why should an organization be concerned with promoting goodprivacy practices? In addition to federal laws and regulations, sitesthat ignore privacy concerns of users may suffer in terms of lessbusiness. Sheehan and Hoy [10] found that as the number ofprivacy concerns increased, the number of registrations at a sitedecreased, incomplete information submissions increased, andcomplaints along with requests for removal from list-servers alsoincreased. Therefore, it is in an organization’s best interest tomaintain a positive relationship with their customers, whichincludes addressing users’ privacy concerns.
Some of these concerns can be ascertained based on the customsettings selected in Privacy Bird by participants in the presentstudy. Participants seemed to be motivated not only by the securityof their financial and personal information, for which breaches canlead to identity theft and credit card fraud, but also by nuisanceconcerns. A common nuisance concern was when a site would notallow a visitor to remove themselves from marketing/mailing lists.Participants were the least apprehensive about a site collectinghabit, behavioral, and preference data about them. As a result, inexchange for services or products, this type of information is easilyacquired. Earp and Baumer [11] revealed that the information mostreadily divulged by users includes gender and age, whereas theinformation least likely to be shared online is their Social Securitynumber [2]. It appears that not only does the type of informationexchanged concern a user but also the type of site itself.
The effects of web site type are in line with the purposes of theindividual web sites. Participants indicated a greater willingness toprovide financial information to financial web sites than to socialnetworking sites. Even though familiarity was rated as fairly lowfor the majority of the sites, the type of site affected participants’trust ratings prior to the participants having visited the sitesthemselves.
The ratings made after participants explored the sites showedthat the color of the Privacy Bird for each site had a significantinfluence on participants’ perceptions of trust, willingness to givefinancial information, and feelings of privacy violation. Aspredicted, trust ratings and willingness to give financial informa-tion decreased significantly from the pre-experimental question-naire to the experimental questionnaire for web sites that
produced a red bird. In contrast, for sites that produced greenand yellow birds, the ratings increased or remained the same. Thisoutcome demonstrates a clear influence of Privacy Bird on users’perceptions of web sites and whether they would interact withthem. In addition to producing a reduction in web site privacyconfidence, red birds resulted in significant perceptions that auser’s privacy would be violated, and, in turn, significantly lesswillingness to make a purchase from these web sites. Interestingly,sites that yielded yellow birds, indicating that there was no postedmachine-readable privacy policy, were interpreted as sites that didnot violate the participants’ privacy preferences. Although, half ofthe users did refer to the actual privacy policy for sites that yieldedthe yellow bird, because the yellow bird can give the impressionthat a site is ‘‘safe’’ from privacy violations, it may be more usefulfor Privacy Bird to give a stronger caution designation to thosesites.
In summary, Privacy Bird’s color did influence participants’ratings of trust and willingness to give financial information,among other things. Thus, Privacy Bird seems to be a good tool forwarning users against privacy violations. However, it should bekept in mind that the data were ratings obtained from collegestudents knowledgeable of computers in a laboratory settingwhere privacy was emphasized. Whether Privacy Bird will showevidence of a similar benefit on behavior of a broader range of usersin everyday use of the web is a matter that requires furtherinvestigation. However, since college students are a part of the e-commerce population, the basic findings of this study shouldgeneralize to a broader population of e-commerce users. Thesefindings suggest that e-commerce sites should address the privacyconcerns of users and make salient the cues that inform users thattheir privacy is protected.
An additional concern is that Proctor et al. [1] found that usersshowed confusion about the information that Privacy Bird couldprovide them. When users were asked to set Privacy Bird to checkfor specific privacy violations, users set the interface correctly onlyabout 50% of the time to check for privacy concerns that could beaccommodated by Privacy Bird. Of more concern, 50% of usersoften thought that they had set Privacy Bird to warn them ofprivacy violations that could not be checked by the agent. It isnecessary to further investigate the interfaces of user agents suchas Privacy Bird and make adjustments to those interfaces in orderto improve the user experience and facilitate widespread adoptionof the agents.
Much more needs to be done in the way of ensuring both thatorganizations adhere to good privacy practices and that individualusers are equipped to make informed choices as to whether toprovide their personal information to a site based on the site’sprivacy policy. This can be accomplished in part through guidelinesand regulations, and the development of technologies such asPrivacy Bird. However, because assuring privacy requires complexinteractions of many people with computer networks and systems,human factors and usability evaluations must be an important partof any research program intended to address privacy-related issues[12].
Acknowledgements
This research was supported in part by NSF ITR Cyber Trustgrant #0430274 to Purdue University and a CSULB Scholarly andCreative Activities Committee Award to Kim-Phuong L. Vu. Wethank Stephanie Reinke for help with testing participants.
References
[1] R.W. Proctor, M.A. Ali, K.-P.L. Vu, Examining usability of web privacy policies,International Journal of Human–Computer Interaction 24 (2008) 307–328.
K.-P. Vu et al. / Computers in Industry 61 (2010) 311–317 317
[2] W.F. Adkinson, J.A. Eisenach, T.M. Lenard, Privacy Online: A Report on theInformation Practices and Policies of Commercial Web Sites, The Progress andFreedom Foundation, Special Report, Washington, DC, 2002.
[3] M.A. Graber, D.M. D’Alessandro, J. Johnson-West, Reading level of privacypolicies on Internet health web sites, Journal of Family Practice 51 (2002)642–645.
[4] C. Jensen, J. Potts, Privacy policies as decision-making tools: an evaluationof online privacy notices, in: Proceedings of the SIGCHI Conference onHuman Factors in Computing Systems, ACM Press, New York, 2004 , pp.471–478.
[5] K.-P.L. Vu, F.P. Garcia, D. Nelson, J. Sulaitis, B. Creekmur, V. Chambers, R.W. Proctor,Examining user privacy practices while shopping online: what are userslooking for? in: M.J. Smith, G. Salvendy (Eds.), Human Interface. Part II. HCII2007, Lecture Notes in Computer Science 4558, Springer-Verlag, Berlin, 2007, pp.792–801.
[6] K.-P.L. Vu, V. Chambers, F.P. Garcia, B. Creekmur, J. Sulaitis, D. Nelson, R. Pierce,R.W. Proctor, How users read and comprehend privacy policies, in: M.J. Smith, G.Salvendy (Eds.), Human Interface. Part II. HCII 2007, Lecture Notes in ComputerScience 4558, Springer-Verlag, Berlin, 2007, pp. 802–811.
[7] S. Byers, L.F. Cranor, D. Kormann, P. McDaniel, Searching for privacy: design andimplementation of a P3P-enabled search engine, in: Proceedings of the 2004Workshop on Privacy Enhancing Technologies (PET2004), Toronto, Canada, May26–28, 2004.
[8] L.F. Cranor, M. Arjula, P. Guduru, Use of a P3P user agent by early adopters, in:Proceedings of the ACM Workshop on Privacy in the Electronic Society, ACM, NewYork, 2002, pp. 1–10.
[9] R.W. Proctor, K.-P.L. Vu, A multimethod approach to examining usability of Webprivacy polices and user agents for specifying privacy preferences, BehaviorResearch Methods 39 (2007) 205–211.
[10] K.V. Sheehan, M.G. Hoy, Flaming, complaining, abstaining: how online usersrespond to privacy concerns, Journal of Advertising 28 (3) (1999) 37–51.
[11] J.B. Earp, D. Baumer, Innovative web use to learn about consumer behavior andonline privacy, Communications of the ACM 46 (4) (2003) 81–83.
[12] R.W. Proctor, E.E. Schultz, K.-P.L. Vu, Human factors in information security andprivacy, in: J. Gupta, S. Sharma (Eds.), Handbook of Research on InformationSecurity and Assurance, Idea Group, Inc., Hershey, PA, 2009, pp. 402–414.
Kim Vu is an Associate Professor of Psychology at
California State University Long Beach. She is also
associate director of the Center for Usability in Design
and Accessibility.
Vanessa Chambers is a graduate student in the
Psychology Department’s Master’s of Science- Human
Factors program at California State University Long
Beach.
Beth Creekmur is a graduate student in the Psychology
Department’s Master’s of Arts-Research program at
California State University Long Beach.
Dongbin (Tobin) Cho is a Ph.D. student in cognitive
area of the Department of Psychological Sciences at
Purdue University. He received a Master’s degree from
Purdue in 2009.
Robert Proctor is a distinguished Professor of Psycho-
logical Sciences at Purdue University with a courtesy
appointment in Industrial Engineering. He is also a