Examining the Reinforcing Properties of Making Sense: A Preliminary Investigation

The Psychological Record, 2012, 62, 599–622

This research was partially supported by a grant funded by the Student Resources Allocation Committee of the Graduate and Professional Student Association at the University of New Mexico.

We would like to thank Lauren Bullard, Seamus Gentz, Beth Leitman, and Dustin Truitt for their assistance in data collection and Josh Tybur for his assistance with data analyses.

Paul M. Guinther is now at Portland Psychotherapy in Portland, OR.Correspondence regarding this article should be addressed to Alisha M. Wray, North Florida/South Georgia

Veteran’s Health System, 1601 SW Archer Rd., Gainseville, FL 32608-1197. E-mail: [email protected]

EXAMINING THE REINFORCING PROPERTIES OF MAKING SENSE: A PRELIMINARY INVESTIGATION

Alisha M. Wray, Michael J. Dougher, Derek A. Hamilton, and Paul M. Guinther

University of New Mexico

Acceptance and commitment therapy asserts that in clinical problems such as rumination and depression, making sense continues despite accompanying aversive consequences, because sense- making is reinforcing, particularly when it leads to experiential avoidance. The following series of experiments aimed to provide preliminary empirical evidence for this hypothesis by comparing college students’ preferences for a solvable laboratory task with response- contingent reinforcement to a formally similar but unsolvable task, on which equal amounts of reinforcement were presented independent of performance. When asked to choose, participants reliably preferred solvable or neutral tasks over unsolvable tasks according to self- report as well as concurrent and forced- choice behavioral procedures. These studies also aimed to investigate whether individuals preferred the solvable condition because it was positively reinforcing or because avoiding the unsolvable condition was negatively rein-forcing. Results are consistent with the theory that making sense may continue despite adverse consequences because of experiential avoidance. Key words: Acceptance and commitment therapy (ACT), making sense, making meaning, reason- giving, coherence, rumination, depression

Many researchers have devoted attention to the human tendency to find meaning or make sense of events in the environment. This tendency to evaluate, analyze, and under-stand has been given a variety of terms, including forming a self- narrative (Gergen & Gergen, 1988; Pennebaker & Seagal, 1999), making/finding meaning (Janoff- Bulman & Frieze, 1983), developing a sense of coherence (Antonovsky, 1979), post traumatic growth (Calhoun & Tedeschi, 2000), and making sense or reason giving (Hayes, Barnes- Holmes, & Roche, 2001; Hayes, Strosahl, & Wilson, 1999). Many other, related cognitive behaviors such as self- focus (Carver & Scheier, 1981; Duval & Wicklund, 1972; Pyszczynski & Greenberg, 1986), repetitive thought (Watkins, 2008), and rumination (Martin & Tesser, 1996; Nolen- Hoeksema, 1991; Papageorgiou & Siegle, 2003; Papageorgiou & Wells, 2001, 2003) can also be conceptualized as sense- making because they share a commonality. Specifically, all of these terms refer to cognitive or verbal behaviors that are evoked in problem situations, which Skinner (1953) described as situations in which reinforcement

600 WRAY ET AL.

may be available but there is “no behavior immediately available” to the organism that can produce the desired outcome or reinforcement (Skinner, 1953, p. 246). Functionally, making sense, as well as these related behaviors, can be conceptualized as any operant behavior that is evoked in contexts where no effective behavior is immediately available and is established and maintained by reinforcement.

A sense- making repertoire is likely reinforced because it allows individuals to effec-tively solve problems. The behavior of generating a narrative that accurately, coherently, and contingently describes relations among events often allows more effective action. For example, individuals who can describe the relationship between an empty gas tank and the operation of their automobile may likely be able to use this narrative to operate their vehi-cle more effectively. In fact, research shows that an organism’s ability to contingently impact its environment is critical in both positive and negative reinforcement (for a review, see Mineka & Henderson, 1985). For example, individuals prefer to work for reinforce-ment, rather than be delivered reinforcement that is not contingent on their behavior. As such, conditions and behaviors that result in effective action are likely reinforcing.

Beyond immediate effective action, Gergen and Gergen (1988) suggested that coher-ent self- narratives are essential for establishing credibility in the social community, main-taining relationships, and obtaining social reinforcement. Consider, for example, a child who has recently received a high grade on a spelling test. The child’s parents may ask her, “How were you able to get such a good grade?” More reinforcement is likely to be avail-able for answers that satisfy multiple social criteria. That is, answers that are grammati-cally correct, make sense, and are consistent with the parents’ values (e.g., “I worked hard and studied”) are more likely to be reinforced than nonsensical answers (e.g., “The jungle ate my book”) or answers that are inconsistent with parental values (e.g., “Oh, I was just lucky”). Only verbal behavior that is sensible based on what the social community deems to be appropriate will be reliably reinforced.

Sense- making occurs and is reinforced in multiple contexts. As a behavior is rein-forced in multiple contexts, it can become a generalized operant class (Gerwirtz & Stengle, 1968). A commonly cited example of this type of operant class is generalized imitation, where after repeated reinforcement for imitating specific models, the act of imitating itself becomes a generalized class that occurs in the presence of novel models or in novel con-texts. In a similar way, making sense can function as a generalized operant class that emerges in a variety of contexts and eventually is maintained, at least in part, without external reinforcement. That is, it can become generally reinforcing (Hayes et al., 1999, 2001) because of a history of reinforcement for engaging in sense- making.

When a behavior becomes generally reinforcing, it can also acquire “inherently,” or automatically, rewarding properties in as much as engaging in the behavior itself can become rewarding. Although Skinner did not provide an explicit definition, he often used the term automatic reinforcement to refer to behaviors that are reinforced without the delivery of reinforcement from another person, which is essential in an analysis of com-plex verbal behavior such as problem solving (Vaughan & Michael, 1982). Moreover, auto-matic reinforcement has been used synonymously with the terms intrinsic reinforcement and self- reinforcement (Vaughan & Michael, 1982), which all suggest that the behavior increases and is maintained, at least in part, by the inherently rewarding aspects of engag-ing in the behavior (e.g., it is fun, it is comforting). This is not to say that the automatically reinforced behavior did not acquire its “inherent” reinforcing properties through a learning history (Hayes et al., 1999, 2001; Sundberg & Michael, 2001) or that it cannot be subjected to experimental analysis (Vaughan & Michael, 1982). However, the main point is that, in addition to external reinforcement, making sense may become reinforcing to the organism in and of itself, which may contribute to its pervasiveness.

Consistent with this idea, there is evidence that sense- making emerges in the absence of external reinforcement (Bruner & Revusky, 1961; Chapman & Chapman, 1967; Golding & Rorer, 1972; Holyoak & Simon, 1999; Kelley, 1972, 1973; Peterson & Seligman, 1984; Skinner, 1936, 1953; Starr & Katlin, 1969; Ward & Jenkins, 1965). For example, Peterson

601MAKING SENSE

and Seligman (1984) asked participants to describe the two worst events they had experi-enced within the last year, with the only other instructions being to limit their descriptions to 250 to 300 words. Results indicated that participants spontaneously offered causal explana-tions of these experiences without any prompting to do so. Similarly, Skinner (1936, 1953) reported that when individuals were asked to report what they heard when listening to ambiguous auditory stimuli (e.g., “ooh, uh, ooh, ooh, ah”), they often reported hearing mean-ingful stimuli, such as words used when conversing with the experimenter or words describ-ing hunger or fatigue. Additionally, evidence suggests that individuals spontaneously perform in consistent and coherent ways in the absence of contingent, external reinforcement for doing so (Harrison & Green, 1990). These findings suggest that individuals spontaneously attempt to make sense of their environments, even when not specifically required to do so.

Making Sense and Psychological ProblemsBecause making sense may increase an individual’s degree of accurate prediction and

effectiveness (for a review, see Mineka & Henderson, 1985), previous conceptualizations have viewed attempts to make sense about personal problems as an adaptive verbal pro-cess. Finding meaning in or making sense of aversive or traumatic events has been reported to lead to psychological benefits, including reducing the likelihood of ensuing depression or posttraumatic stress disorder (e.g., Mendola, Tennen, Affleck, McCann, & Fitzgerald, 1990). Mendola and colleagues (1990), for example, found that infertile women who could identify a positive outcome that came from their infertility, such as becoming closer to their partner, were buffered against subsequent psychological problems. In addition, making sense has been associated with psychological well- being (Pals, 2006), self- esteem (Bird & Reese, 2006; McAdams, Reynolds, Lewis, Patten, & Bowman, 2001), and increased health benefits, such as improved immune system functioning (e.g., Affleck, Tennen, Croog, & Levine, 1987; Bower, Kemeny, Taylor, & Fahey, 1998). Moreover, Pennebaker and Seagal (1999) found that a variety of populations (e.g., maximum- security prisoners, distressed crime victims, chronic pain sufferers) benefit from writing about aversive events, which facilitates sense- making.

Despite the obvious benefits of making sense, a growing body of literature suggests that sense- making about psychological problems is not always adaptive (for a review, see Watkins, 2008) and may actually produce negative or aversive outcomes (Borkovec, Robinson, Pruzinsky, & Depree, 1983; Carver & Scheier, 1981; Duval & Wicklund, 1972; Martin & Tesser, 1996; Nolen- Hoeksema, 1991; Papageorgiou & Siegle, 2003; Papageorgiou & Wells, 2001, 2003; Pyszczynski & Greenberg, 1986; Watkins, 2008). For example, consider the same child previously provided as an example, who has now failed a test. When asked for the cause by her parents, she now attributes her failure to not studying enough, which is perhaps punished momentarily (e.g., stern talking to, imposed restric-tions), but it is a socially appropriate causal answer. However, another child might attribute the failure to being unintelligent. This answer may satisfy social contingencies (i.e., is sensible, coherent, etc.), but it may increase social criticism and negative affect (Wray, Freund, & Dougher, 2009). Research has shown that this style of sense- making can be quite problematic. Ingram (1990), for example, reported several studies that demonstrate an association between sense- making and a variety of clinical problems, including anxiety (Buss, 1980; Carver & Scheier, 1986; Sarason, 1975, 1986; Wine, 1971, 1982), alcohol abuse (Hull, 1981), and, most frequently, depression (Lewinsohn, Hoberman, & Hautzinger, 1985; Musson & Alloy, 1988; Pyszczynski & Greenberg, 1987).

Studies also show that experimentally induced rumination increases depressed mood (e.g., Park, Goodyer, & Teasdale, 2004) and reduces engagement in positively reinforcing activities (e.g., Brockner & Hulton, 1978; Lyubomirsky & Nolen- Hoeksema, 1993). Similarly, a growing body of literature finds that sense- making about one’s depression can interfere with psychological treatments. For example, Addis and Jacobson (1996) demon-strated that sense- making is correlated with poorer outcomes in behavioral activation

602 WRAY ET AL.

treatment and that clients who attribute their depression to childhood and relationship issues respond more poorly to both cognitive and behavioral treatments. Furthermore, making sense tends to decrease with psychological treatment (Natale, Dahlberg, & Jaffe, 1978). These findings indicate that increased sense- making may be not only ineffective in solving psychological problems but also quite costly.

These findings have led researchers to wonder why sense- making would continue as an avenue for solving psychological problems when it can lead to negative outcomes. Some have suggested (e.g., Hayes et al., 1999, 2001) that because sense- making has been rein-forced so extensively and acquires automatically reinforcing functions, it can become pervasive. Said another way, maladaptive sense- making, such as rumination and worry that are observed in clinical problems, may be conceptualized as a behavioral excess, or a behavior occurring in an inappropriate context, that is maintained by generalized and auto-matic reinforcement, even when it increases negative affect.

Making Sense and Acceptance and Commitment TherapyAcceptance and commitment therapy (ACT; Hayes et al., 1999) is a behaviorally based

psychological intervention that incorporates mindfulness and acceptance strategies to increase psychological flexibility and subsequently valued living. This approach is in direct contrast to traditional cognitive- behavioral treatment approaches (CBT) that aim to reduce psychological symptoms by not only implementing behavioral changes but also altering cognitions (Beck, 1967; Beck, Rush, Shaw, & Emery, 1979; Ellis, 1962, 1990). Although ACT is a relatively new evidence- based treatment, it is quickly acquiring empirical support across a wide array of clinical problems (Hayes, Luoma, Bond, Masuda, & Lillis, 2006).

Making sense plays a critical role in ACT. Specifically, Hayes and colleagues (Hayes et al., 1999, 2001) assert that while making sense may be helpful in solving many problems in the environment, it is not helpful—and in fact is problematic—for solving psychological problems, particularly when solving the psychological problem involves attempts to control emotional experiences (Hayes & Wilson, 1994). A theoretical tenet of ACT is that sense- making generalizes from external problem situations to psychological problem situations in an attempt to identify the causes of emotional distress so as to escape or avoid it. However, another tenet of ACT is that experiential avoidance is at the core of a variety of clinical problems, suggesting that sense- making or reason giving, when used to avoid emotional distress, may actually exacerbate psychological problems (e.g., Hayes et al., 2004).

To illustrate, consider the second child in the previous example. She is now 16 years old and has repeatedly accounted for various failures, which cause considerable psycho-logical distress, by attributing her mistakes to a characterological flaw (e.g., unintelli-gence). These attributions are meant to help identify the cause of her distress (failures) so that she may resolve them to escape the aversive feelings accompanying them. However, this sense- making also carries a tremendous cost, in as much as it exacerbates the negative affect associated with the failure and it does not actually help her perform better on her tests, to avoid future failures. In fact, she may study less because she may believe studying will not help her because she is unintelligent. These problematic sense- making behaviors may be accompanied by other behaviors, such as cognitive bias that can exacerbate clini-cal problems by increasing the salience of negative affect, environmental events, and self- talk (see Wray et al., 2009, for a detailed discussion of this topic).

As mentioned earlier, the acquired generally and automatically reinforcing functions of making sense may explain why it occurs in the presence of psychological distress and despite its ineffectiveness (Addis & Carpenter, 1999; Addis & Jacobson, 1996) and accompanying aversive consequences (Martin & Tesser, 1996; Nolen- Hoeksema, 1991). Although this assertion seems plausible, it has not yet been shown empirically that making sense can, in fact, function as a reinforcer. The dearth of empirical research in this important and clini-cally applicable area may be due to the challenges that arise when studying complex verbal behavior. However, one way to preliminarily investigate this complex question is to utilize a basic, translational study that allows for many variables to be controlled while allowing the

603MAKING SENSE

relevant variables to be isolated. Specifically, in this study, we aim to use a basic, laboratory design to compare participants’ preference for a solvable laboratory task (one in which con-tingent feedback is presented based upon consistent contingencies) to a formally similar but unsolvable task (one in which noncontingent feedback and inconsistent contingencies are presented), with the amount of positive feedback in the two tasks held constant. If sense- making is reinforcing, then participants should prefer the solvable task.

Experiment 1

MethodParticipants. Nineteen volunteers were recruited from the subject pool of a local

university’s psychology courses, through a recruitment Web site. The data for 17 partici-pants were included in the final analyses. Two participants’ data were excluded: one due to technical difficulties and one for lack of attention to the experimental tasks. Sixteen of the 17 participants were female, and their ages ranged from 18–41 years, with a median age of 19 years. Participants self- identified their ethnicities as follows: 53% Caucasian/White, 41% Hispanic, and 6% other.

All participants received course credit for participating in the study. Participants were informed that they would be participating in a “study that is investigating the way that individuals think during cognitive tasks and the choices that they make based on what they have learned.” Informed consent was obtained, and accepted Institutional Review Board procedures were followed.

Setting, apparatus, and materials. Participants were seated at a table with a com-puter located in an adjacent room that presented stimuli and recorded data. Stimuli con-sisted of 36 abstract stimuli and 12 three- letter nonsense syllables (example stimuli are presented in Figure 1). All stimuli were squares containing colored designs against a black background.1 The stimuli used in this experiment and the software that controlled it were created specifically for this experiment and are available upon request.

ZAK HEF LUGDIT TOB YIMNUK MIF WECSAJ VEP GOL

Figure 1. Six of the 36 abstract figures and the 12 nonsense syllables.

1 All figures are printed in black and white, but they were created and shown to participants in color. Figures in color are available from the first author upon request.

604 WRAY ET AL.

Design. The study used a within- subjects design, in which all participants took part in all experimental conditions.

General procedure. All participants were tested individually. Upon completion of the statement of informed consent and a demographics questionnaire, the experimenter read instructions2 that oriented participants to the task and then initiated the experiment.

The first experiment involved two conditions consisting of two match- to- sample (MTS) tasks. MTS tasks were chosen because they are commonly used in behavioral research of verbal processes, such as derived relational responding research. In the first MTS task (S), the task was solvable, meaning that contingencies remained consistent throughout the task and that feedback was provided contingent on participants’ perfor-mance. Positive feedback (“Correct”) was delivered contingent on each correct selection, whereas negative feedback (“Wrong”) was delivered contingent on each incorrect selec-tion. In the second MTS task (U), the task was unsolvable, meaning that there was no relationship between antecedent stimuli and response feedback. Specifically, positive and negative feedback was yoked to the feedback that participants had received during their S task performance. That is, the feedback that participants received during the U task was independent of their performance on the U task and was instead delivered at the same rate and in the same sequence as the feedback that participants received during the S task. The U task was completed after participants completed as many trials on the U task as it took them to complete the S task. Subsequent to the end of the U task, participants were asked to choose which, if either, of the previous two conditions they would want to repeat in the next phase of the study. Participants’ self- reported preferences were the main dependent variable.

Solvable MTS task. In the first condition, the solvable MTS task was designed to establish three contextually controlled, conditional three- member stimulus classes. A complex MTS task was chosen to ensure that participants had a sufficient number of trials to interact with each task, which would presumably improve participants’ ability to discern the tasks’ solvability.

The nonsense syllables served as contextual stimuli, and the abstract figures served as conditional (sample) and discriminative (comparison) stimuli. The procedure required participants to learn which comparisons to select in the presence of each of three samples and according to three contextual stimuli. Thus, correct selections depended on both the contextual stimulus and the sample presented on a given trial (see Figure 2 for specific relations trained). For example, in the presence of the contextual stimulus ZAK, partici-pants were trained to select the comparisons with the same numeric designation as the samples. Thus, given the nonsense syllable ZAK, sample A1, and comparisons B1, B2, and B3, B1 was the correct selection. However, in the presence of the contextual stimulus HEF, a different set of relations was trained. For example, in the presence of HEF, sample A1, and comparisons B1, B2, and B3, B2 was the correct comparison. A typical MTS trial is presented in Figure 3.

On each trial, participants chose the left, middle, or right stimulus by pressing the left, middle, or right arrow keys, respectively. In total, there were 18 learned relations in each condition, each of which was represented by six trial types. A total of 18 relations were randomly presented until participants responded correctly on five of seven (approximately 70% correct) presentations of any relation. Participants were only required to reach a 70% correct criterion, as opposed to a higher (e.g., 90%) criterion, because it was thought that participants would be unable to differentiate between the solvable task and the unsolvable,

2 “You will be participating in a study that is investigating how people learn different relationships that vary under different conditions. We want you to figure out how things go together as best you can. Sometimes the relationships will be easy to determine, and other times they will be more difficult. Sometimes the relationships will be more obvious than others. Always use the computer feedback as your guide. Your advancement to the next phase of the experiment depends upon you accurately learning the relationships, so try your best at all times. Do not make your choices based on the pictures you like the best or which pictures you think are the prettiest. Only pay attention to the associations and the feedback when choosing. Do you have any questions?”

605MAKING SENSE

yoked task if the density of positive feedback reached higher percentage levels. Once par-ticipants reached criterion, trial types of those relations were eliminated from the training cycle until participants met criterion on all relations.

ZAK HEF LUG

A1 A2 A3 A1 A2 A3 A1 A2 A3

B1 B2 B3 B2 B3 B1 B3 B1 B2

C1 C2 C3 C3 C1 C2 C2 C3 C1

DIT YIM TOB

randomly presented randomly presented randomly presented

Figure 2. Example of contextual stimuli and reinforced relations in the S task (top panel) and the U task (bottom panel). Note that the alphanumeric designations (e.g., A1, B2) refer to specific abstract figures. In the U task, the sample and comparison stimuli were randomly selected on each trial.

Unsolvable MTS task. In the second condition, the unsolvable MTS task was identical to solvable MTS task, except that 12 new stimuli were used as samples and comparisons and feedback was yoked to the S task, independent of participants’ performance.3 The yoking procedure required the S task to be presented first, which prevented counterbalanc-ing of the order of the tasks. However, the yoking procedure was used to ensure that any observed differences in preference were the result of the manipulated independent variable (i.e., solvability) rather than other confounding variables (e.g., density of reinforcement). Whereas the S task terminated when participants met the training criterion of five out of seven correct responses on all trial types, the U task terminated after participants received the same number of trials as was required to meet criterion in the S task.

After completing the U task, participants were asked by the experimenter to choose the condition they would prefer to work on for an additional 30 min. Participants were given the following three options: Condition 1 (S), Condition 2 (U), or No Preference. After participants gave their responses, the experiment was terminated, and participants were debriefed and dismissed.

ResultsThe mean number of trials required for participants to meet criterion during the S task

was 209.29 (SD = 116.06). Participant 1 was a clear outlier, requiring 601 trials to meet criterion. Removing her data from the analysis resulted in more representative means and standard deviations (S task: M = 184.81, SD = 59.15). There is no relevant data to report for the U task, as it was yoked to performance on the S task. Results for each individual par-ticipant are available upon request.

Participants’ choice of conditions subsequent to the completion of both conditions was the primary dependent variable. Eleven of the seventeen participants (65%) chose the S task, whereas only four participants (23%) chose the U task. Two participants (12%) reported no preference. A chi square analysis revealed the selection pattern to be signifi-cant, χ2 (2, n = 17) = .019. Because the low number of participants in two of the cells threatened the reliability of the χ2 statistic, additional analyses were performed.

A binomial probability can be determined if participant preferences are separated into two categories. Since there are three total categories, S, U, or no preference, participant preferences were separated into the following two categories for the sake of the binomial

3 A second version of this study was conducted where the S task was identical to that proposed in Experiment 1, but the U task randomly presented correct feedback on 90% of the trials in contrast to the yoked 70% presented here. This experimental manipulation resulted in the majority of participants being unable to discern between the two tasks and is therefore not presented here. Data is available from the first author upon request.

606 WRAY ET AL.

distribution analysis: (1) participants who preferred the S task (expected probability = .33) and (2) those who preferred either of the other two options (expected probability = .67). The binomial probability, then, that 11 of 17 participants would prefer the solvable condi-tion by chance alone is .014.

Contextual Stimulus

ZEK

Sample StimulusA1

1-Second Delay

Comparison Stimuli

B1 B2 B3

Figure 3. Example of a match- to- sample trial used in this experiment.

DiscussionThe purpose of this study was to test the idea that sense- making is reinforcing in a

laboratory setting. The large majority of participants in this study preferred experimental tasks that were solvable. While it is possible, and likely, that participants preferred the solv-able condition because it provided accurate, well- timed, and contingent positive feedback, it is also reasonable to conclude that the solvability of the task (which might be well opera-tionalized by the three preceding characteristics) was influencing participant preference. Regardless, these results provide preliminary empirical support that making sense can be reinforcing, which is relevant to the growing body of literature suggesting that making sense can be maladaptive under certain circumstances. If it is at times maladaptive, how-ever, then a legitimate question would be why it continues. Some researchers (Hayes et al., 1999, 2001; Martin & Tesser, 1996; Nolen- Hoeksema, 1991) have suggested that making sense is generally reinforcing, which would explain why individuals would engage in sense- making even in situations where it may not be useful and may even be aversive (Addis & Carpenter, 1999; Addis & Jacobson, 1996; Hayes et al., 1999, 2001). While this idea is a foundational assumption of ACT, evidence to this point has been only indirect and speculative. Thus, this study fills an important theoretical gap in the current literature.

While this study provides support for making sense as a reinforcer, one limitation of the present study is that it did not determine why making sense is reinforcing. That is, it is

607MAKING SENSE

unclear whether individuals preferred the solvable condition because it was positively reinforcing or because avoiding the unsolvable condition was negatively reinforcing. There is some evidence to support both explanations, and both would be consistent with the assumption that making sense may serve experiential avoidance. To the extent that making sense has been useful in some contexts in terms of identifying the causes of a problem, it may acquire generalized reinforcing properties and naturally occur in contexts where one is trying to identify and remove the causes of emotional distress. In addition, there is evi-dence to suggest that the ambiguity resulting from an unresolved problem is aversive for most individuals, particularly when in conjunction with an aversive event, and the removal of the conditions causing this ambiguity is negatively reinforcing (Averill, 1973; Burger & Arkin, 1980; Glass & Singer, 1972; Sosnowski, 1983, 1988).

A distinction between these two reinforcing functions has important theoretical as well as clinical implications. For example, if a client engages in excessive sense- making to reduce his or her anxiety related to an inability to control his or her college- age child’s behavior (e.g., dating partners, substance use, study habits), exercises increasing his or her ability to tolerate anxiety (e.g., mindfulness) might be implicated. On the other hand, if another client engages in excessive sense- making in order to obtain social approval, which is actually hindering his or her ability to respond contingently to the interaction, communi-cation skills (e.g., active listening) might be highlighted. Given these implications, the first aim of the second study is to establish whether the preferences observed in the previous study were due to the positive reinforcement associated with sense- making or the negative reinforcement associated with avoiding unsolvable situations.

One way to evaluate the role of multiple reinforcing functions is to introduce a neu-tral task that provides individuals with an alternative to both the solvable and unsolvable conditions. Differential preference for the solvable task or the neutral task over the unsolvable task would indicate that making sense is reinforcing and that conditions that prevent sense- making are aversive. Further, differential preference for the solvable over the unsolvable and neutral tasks would indicate positively reinforcing functions. Based on the literature suggesting that unsolvable conditions are experienced as aversive, it was expected that individuals would prefer both solvable and neutral conditions to unsolvable conditions. Additionally, because solvable conditions allow individuals to access both reinforcing functions simultaneously, it was anticipated that solvable condi-tions would be preferred most, followed by neutral and unsolvable conditions, respectively.

Finally, a second limitation of the first experiment was that participant preferences were based solely on self- report. Although it is reasonable to assume in this case that these reports were in line with participants’ actual behavioral preferences, it is preferable to demonstrate that the contingent availability of sense- making tasks would actually function to reinforce other operants. In order to address this limitation, the second study aims to replicate these earlier findings using a concurrent chaining method, which is a commonly used preference task that uses a behavioral measure of preference.

Experiment 2

MethodParticipants. Fifteen volunteer participants were recruited from the university psy-

chology department subject pool through a recruitment Web site. No participants were excluded. Two of the 15 participants were male, and participant ages ranged from 18–47 years, with a median age of 19. Participants self- identified their ethnicities as follows: 53% Caucasian/White, 33% Hispanic, 7% Black/African American, and 7% Native American. As in Experiment 1, participants were informed about the study and consent was granted, participants received course credit, and Institutional Review Board procedures were followed.

608 WRAY ET AL.

Setting, apparatus, and materials. The setting and apparatus were the same as those in Experiment 1. Comparison and sample stimuli consisted of 36 visual stimuli that were composed of 3.56 in2 pictures of women and men with common names. All of the visual stimuli were obtained from publicly accessible Internet sites and were modified to be the same in size and resolution. No differences in preferences have been observed in our previ-ous work when using abstract or social stimuli (e.g., Wray & Dougher, 2009; Wray, Dougher, & Bullard, 2008), and it was thought that the use of socially relevant stimuli might help improve participant interest in the task. However, to ensure participant prefer-ences were not based on systematic preference for the specific stimuli, all visual stimuli were randomly assigned to each task for each participant. In addition, all stimuli were rated for attractiveness by seven undergraduate research assistants on a 5-point Likert- type scale, where 1 = very unattractive, 2 = unattractive, 3 = neither unattractive nor attrac-tive, 4 = attractive, and 5 = very attractive. The mean attractiveness rating was calculated (M = 3.08) and was considered to be within a neutral range. All included stimuli had an average rating within 1 point of the overall mean (range of mean ratings = 2.08–4.08). Examples of the 36 social stimuli are presented in Figure 4.

Design. Experiment 2 used a within- subjects design, in which all of the participants participated in all of the experimental conditions.

General procedure. The experiment consisted of two phases. In the first phase, par-ticipants were presented with four conditions: (1) a solvable, conditional, arbitrary MTS task (S), (2) a neutral task (N1), (3) an unsolvable, arbitrary MTS task (U), and (4) another neutral task (N2).

After reading all of the Phase 1 instructions4 that oriented participants to the task, participants were presented with a task- selection screen (see Figure 5), in which the A was illuminated. Pressing the A key on the keyboard illuminated the H on the task- selection screen, and pressing the corresponding H key granted access to the solvable (S) task. Participants always began with the S task, where correct feedback was delivered contin-gent on performance. After completing 10 trials of the S task, participants were randomly assigned to work on one of the two neutral tasks (N1 or N2). Participants pressed the appro-priate key to gain access to each task (J for N1 and K for N2). After completing 10 trials of the first neutral task, participants were exposed to an unsolvable MTS task (U). After completing 10 U trials, participants were exposed to the other neutral task that was not randomly assigned following S. In review, participants first worked on the S task, then on either the N1 or N2 task, then on the U task, and finally on the alternate N1 or N2 task.

4 “You will be participating in a study that is investigating people’s preferences regarding different kinds of tasks. Starting off, you won’t have much choice about which tasks you will perform, but later in the experiment you will be able to choose freely. To gain access to each type of task, you will first be shown a screen with the letters A and S near the top, and the letters H, J, K, and L near the bottom. You can only pick letters that are illuminated. Picking A will then allow you to pick between H and J, whereas picking S will then allow you to pick between K and L. For now, you will automatically spend some time alternating between the A tasks (H and J), and then you will automatically spend some time alternating between the S tasks (K and L). So just pick the illuminated letters and you’ll be on your way. [Press G for more instructions.]

“The H task and the L task require you to figure out relationships. There are several problems where there is a picture on the top of the screen, and three pictures on the bottom of the screen. Your job is to figure out which pictures on the bottom of the screen go with which pictures on the top of the screen. Figure out how things go together as best you can. At the beginning, you will get feedback as to whether you picked the right picture, and you can then press the space bar to bring up the next problem. Always use the computer feedback as your guide to figure out the right answer. Only pay attention to the associations and the feedback when choosing. Your advancement to the next phase of the experiment depends on you accurately learning the relationships, so try your best at all times. To pick the picture on the left, press 1, to pick the picture in the middle, press 2, and to pick the picture on the right, press 3. [Press G for more instructions.]

“The J task and the K task are similar but only require you to look at a set of pictures and press 1, 2, or 3, for the picture on the bottom that matches the top picture. Your ONLY job in these tasks is to match pictures, then press the space bar to move to the next trial.

“You are free to take breaks at any time—just let the experimenter know. Please ask the experimenter if you have any questions now or at any point during the experiment.”

609MAKING SENSE

Sarah Rachel Lindsey

Erik Dave Ryan

Lilly Casie Callie

Juan Conner John

Figure 4. Visual stimuli used for comparison and sample stimuli in Experiments 1 and 2.

This sequence was repeated until criterion performance (described in the next paragraph) was achieved on the S task.

610 WRAY ET AL.

A S

H J K L

Figure 5. Presentation of task- selection screen presented to participants.

Similar to Experiment 1, in the U task, positive and negative feedback was yoked to the performance in the solvable S task. Because the U task was yoked to the S task, the S task was always presented first. To ensure that participants could not easily detect the yoked feedback, neutral tasks were presented in between the solvable and unsolvable tasks. It was also thought that presenting neutral tasks in between the S and U tasks would reduce any potential order effects on preference. Participants were cycled through alternating presentations of 10 trials of each task type until they met criterion (i.e., five correct out of the last seven trials for an approximate 70% correct criterion) on the S task. Once partici-pants met criterion on the S task, they were cycled through one more set of 10 trials of the remaining tasks (N1 or N2, U, and the alternate N1 or N2) before moving on to Phase 2.

Subsequent to the completion of all of the tasks, they concluded Phase 1 and partici-pants were placed in a choice procedure to determine their preferences for each kind of task, which will be discussed in detail in the next section. Once participants completed the choice phase, the experiment was terminated and participants were debriefed before being dismissed.

Solvable MTS task. The arbitrary MTS task, which permitted sense- making, was intended to establish three 3-member stimulus classes. The social stimuli comprised the stimulus classes. The solvable MTS task procedure was identical to that in Experiment 1, with the following exceptions. First, because the current procedure required participants to complete four tasks, contextual stimuli were removed in order to decrease the amount of time required to complete the experimental procedure. Second, the task was presented in 10-trial increments along with 10-trial increments of the other three tasks rather than in complete, sequential blocks.

Neutral task 1. The S and U tasks were separated by the presentation of neutral tasks, which served to expose participants to tasks that were formally similar to the solvable MTS task but that did not encourage sense- making behavior. The neutral tasks allow investigation of whether participants prefer conditions that are solvable or if they simply prefer to avoid conditions that are unsolvable. During the neutral tasks, participants were exposed to randomly presented stimulus configurations resembling the MTS stimulus arrays, with the exception that nine new stimuli were used.

Participants did not engage in selecting stimuli, as in the S task, for the purpose of solving an arbitrary MTS task, but rather were given a simple identity- matching MTS task. In the identity- matching task, participants were instructed only to match the stimulus on the top of the screen with the identical stimulus on the bottom. An example of this trial type can be seen in Figure 6. Computer feedback was presented to let participants know if they correctly chose the matching stimulus.

611MAKING SENSE

Conner

DaveLaurenConner

Figure 6. Presentation of visual array presented to participants in the neutral tasks.

Unsolvable MTS task. The unsolvable MTS task procedure was identical to the unsolvable MTS task used in Experiment 1, with the exceptions that were previously out-lined for the S task procedures.

Neutral task 2. The second neutral task was identical to the first neutral task with one exception. Participants were not presented with computer feedback, but rather each selec-tion was followed by the presentation of a grey box. Once participants completed all trials in Phase 1, they moved on to Phase 2.

Preference task. In order to assess participants’ differential preferences for the four tasks, they entered a second phase. Participants were given a set of choices based on a concurrent chaining procedure, which is a commonly used method for testing prefer-ence. Concurrent chain procedures use a behavioral measure of preference that requires participants to engage in a choice behavior in order to gain access to an array of reinforc-ers (Catania, 1992). Participants were given a total of 12 concurrent choices among all four conditions. Participants saw the task- selection screen with all four tasks illuminated and selected whichever key matched the task they preferred. Following each selection, participants were presented with 10 additional trials of the chosen task. Participants were asked to complete additional trials for two primary reasons. First, participants were asked to engage in another behavior to gain access to each task, which allows each con-dition to function as a reinforcer. Second, it was thought that participants might more judiciously choose if they were required to engage in the task subsequently. In order to reduce the likelihood that participant preference might be affected by reinforcement during this preference phase, no feedback was presented during any task. Following completion of the 120 trials, participants completed Phase 2 and were debriefed and dismissed.

ResultsThe mean number of trials required for participants to meet the percentage correct

criterion was 62.46 (SD = 16.45). Individual data is available upon request. Conditions N1, N2, and U were yoked to equate the number of trials in the previous phase and thus matched that of the S task. As such, on average, participants completed approximately 240 task trials prior to moving on to Phase 2.

As already mentioned, the dependent variable of the present study was partici-pants’ choice of conditions in the second phase. Because participant preferences did

612 WRAY ET AL.

not show robust within- subject effects (see Appendix A for individual participant pref-erences), participant preferences for S, N, and U tasks were subjected to group analyses.

Prior to comparing participant preferences for S, U, or N tasks, participant preference for the two neutral tasks was investigated with a paired- sample t test. No significant differ-ences emerged between the neutral task with feedback (N1: M = 3.07, SD = 1.33) and the neutral task with no feedback (N2: M = 3.07, SD = 1.49), p = .708, so all neutral data was collapsed into one neutral task by averaging each participant’s preferences for N1 and N2. Hence, the analysis consisted of three levels of a within- subjects variable: S, U, and the average of the two neutral tasks.

Total preferences across participants were compared using a repeated- measures anal-ysis of variance (ANOVA). Because multivariate approaches are less reliant on assump-tions of homogeneity of variances across levels of a within- subjects factor than univariate approaches, a multivariate approach was used (Maxwell & Delaney, 2004), with the mul-tiple variables being the three levels of the within- subjects preference variable.

As detailed in Table 1, the omnibus test revealed significant differences in preference among the three conditions, F = 4.80, p = .028. Two contrasts were subsequently per-formed. First, no significant difference emerged between the S and N tasks, F = 1.30, p = .273, ns. Hence, the final comparison was between the average of the S and N tasks and the U task, which revealed a significant difference between the U task and the combined S and N tasks, F = 5.45, p = .035. In other words, participants preferred the S and N tasks to the U task, but did not show a significant preference between the S and N tasks. Table 1 Participant Preferences in Experiment 2

Analysis F Hypothesis df Error df p

1 – Omnibus 4.80 2 13 .028*

1 – Ψ (S vs. N) 1.30 1 14 .273

1 – Ψ (S/N vs. U) 5.45 1 14 .035*

Note. All tests were conducted using repeated-measures multivariate ANOVAs.*p < .05.

DiscussionResults indicate that participants significantly preferred solvable or neutral conditions

over a condition that was not solvable. These results suggest that unsolvable conditions can be aversive and that it is reinforcing to avoid them. However, there was not a clear prefer-ence for the S task over the N tasks, indicating that although some individuals may find sense- making positively reinforcing, many find simply avoiding unsolvable conditions rewarding.

Although concurrent chaining procedures are a widely used method for investigating preferences among three or more simultaneously presented options, these procedures do not allow for an evaluation of differential preferences when only two of several options are presented for selection. In contrast, forced- choice procedures, in which participants choose between two available options out of several possible options, might highlight differential preferences between each task type. Based on this assumption, it was thought that impos-ing forced choices might improve the ability of participants to distinguish between the solvable and neutral tasks. Additionally, only 12 total concurrent chain choices were offered in Experiment 2, which may have limited the opportunity for comparisons to be made among the conditions. In an attempt to obtain clear preferences between the S and N tasks, minor revisions related to these two observations were made, and additional data were collected in a third experiment.

613MAKING SENSE

Experiment 3

MethodParticipants. Twenty volunteer participants were recruited from the university psy-

chology department subject pool through a recruitment Web site. No participants were excluded. Four of the 20 participants were male. Participant ages ranged from 18–30 years, with a median of 20. Participants’ self- identified ethnicities were as follows: 55% Hispanic, 20% White, 20% other, and 5% Native American. As in Experiments 1 and 2, participants were informed about the study and gave consent, participants received course credit, and Institutional Review Board procedures were followed.

Setting, apparatus, and materials. The setting, apparatus, and materials were iden-tical to those used in Experiment 1, with the following exceptions. A few minor changes were made to the computer program to increase the number of choices in the preference phase and include forced choices rather than concurrent choices. These changes are detailed below.

Design. Experiment 3 used a within- subjects design, in which all of the participants took part in all of the experimental conditions.

General procedure. The general procedure of Experiment 3 was consistent with that of Experiment 2, with the exception of the following minor revisions. In order to increase the stability in within- subject preference, during the preference phase, the number of choices was increased from 12 to 40 total choices. Additionally, in order to highlight the distinction among each comparison type, we imposed forced choices rather than concur-rent choices. Specifically, in the preference phase, participants were given four forced choices and 10 choices of each possible pair.

Phase 1. Phase 1, or the task phase, in Experiment 3 was identical to the task phase in Experiment 2. The only minor exception was that because a concurrent chaining procedure was not being used, the A and S stimuli were removed. As such, participants did not have to first select A or S and only had to press H, J, K, and L to gain access to the four tasks.

Phase 2. In order to assess participants’ differential preference for the four tasks, they entered Phase 2, or the preference task. Participants were given a set of choices based on a forced- choice procedure, which is another commonly used method for testing preference. Participants were given a total of 40 forced choices among pairs of conditions. Participants were given four forced choice combinations: (1) S or U, (2) S or N1, (3) U or N2, and (4) N1 or N2. Participants were given 10 choices of each possible pair—for a total of 40 choices—and three task trials of each task that was selected—for a total of 120 task trials—before completing Phase 2. For each forced- choice trial, participants saw a task- selection screen with only two options illuminated and pressed the key corresponding to the task that they preferred to work on. Following each task selection, participants were presented with three trials of the chosen task before proceeding to the next choice trial. This sequence repeated until participants had made 40 forced choices.

ResultsThe mean number of trials required for participants to meet the percentage correct

criterion was 72.25 (SD = 42.03). Individual data is available upon request. Conditions N1, N2, and U were yoked to equate the number of trials in the previous phase and were the same as the number of S trials.

The dependent variable remained participants’ choice of conditions in Phase 2. The proposed methodological changes did not increase the robustness of the within- subject effect. As such, the results were analyzed similarly to Experiment 2 (see Appendix B for individual participant preferences). However, because there were forced choices, as opposed to concurrent choices, each choice pair was analyzed separately.

614 WRAY ET AL.

Prior to comparing participant preferences for S, U, or N tasks, participant preference for each neutral task was investigated. No significant differences emerged between the neutral task with feedback (N1: M = 12.40, SD = 4.47) and the neutral task without feed-back (N2: M = 10.75, SD = 2.84), t(19) = 1.40, p = .179. As such, both neutral tasks were averaged for all remaining analyses.

As detailed in Table 2, total participant preferences for S, N, and U were subjected to a repeated- measures ANOVA with the omnibus test revealing significant differences in preference among the three conditions, F = 4.35, p = .029. As such, the two follow- up comparisons were conducted. First, no significant difference emerged between the S and N tasks, F = 3.46, p = .078, ns. Second, a significant difference emerged between the U task and the combined S and N tasks, F = 8.32, p =.009. In other words, participants continued to prefer the S and N tasks to the U task, but did not show a significant preference between the S and N tasks. Table 2Participant Preferences in Experiment 3

Experiment F Hypothesis df Error df p

1 – Omnibus 4.35 2 18 .029*

1 – Ψ (S vs. N) 3.46 1 19 .078

1 – Ψ (S/N vs. U) 8.32 1 19 .009**

Note. All tests were conducted using repeated-measures multivariate ANOVAs.*p < .05. **p < .01.

DiscussionExperiment 3 replicated the results obtained in Experiment 2 using a forced- choice

procedure. That is, results indicated that participants equally preferred solvable and neu-tral conditions over the unsolvable condition. Despite increasing the number of choices and imposing a forced- choice procedure, participants continued to show no preference between S and N tasks.

General DiscussionEvidence across multiple psychological research domains indicates that making

sense likely serves reinforcing functions and that these functions may explain why it continues despite accompanying aversive consequences (Hayes et al., 1999; Martin & Tesser, 1996; Nolen- Hoeksema, 1991) and in situations where it may not be useful (Addis & Carpenter, 1999; Addis & Jacobson, 1996). This study provides preliminary evidence that sense- making can indeed function as a reinforcer. Results from three experiments showed that participants consistently preferred a solvable condition over an unsolvable condition despite having equivalent feedback histories with both conditions. This dif-ferential preference was demonstrated using self- reported preference as well as concur-rent and forced- choice behavioral measures. Taken together, these results provide evidence that situations that are solvable are preferred to conditions that cannot be solved.

While Experiment 1 provided support for making sense as a reinforcer, that study did not establish why making sense is preferred. That is, it is unclear whether individuals pre-ferred the solvable condition because it was positively reinforcing or because avoiding the unsolvable condition was negatively reinforcing. The second and third experiments aimed to determine whether making sense functioned as a negative or positive reinforcer within the context of the current experimental designs, as there is evidence for both functions within the literature.

615MAKING SENSE

It is reasonable to assume that differential preference for the solvable over the neutral tasks would indicate that there are additional positively reinforcing aspects of making sense above and beyond the negatively reinforcing functions associated with avoiding unsolvable conditions through selection of the neutral tasks. However, data across the second and third studies highlight that, by and large, there are not strong preferences for solvable over neutral tasks. This suggests that participants may have been more invested in avoiding unsolvable conditions. Said another way, the data presented here indicate that individuals may be less motivated to make sense and more motivated to avoid nonsense.

It is possible that certain reinforcing functions are more powerful for certain individu-als. Although no formal analysis of verbal report data were conducted, it was noted in the postexperimental debriefing that many participants spontaneously offered reasons for their preference. For example, participants who chose neutral tasks often reported that they preferred those tasks because they were “less confusing” or were “obvious,” or because they were tasks that the participant “only had to match.” However, there were also indi-viduals who preferred the solvable tasks. For these individuals, they often reported that the solvable task was “fun to figure out,” was “attainable and not boring,” or “had likable pat-terns,” or they felt like they “did a good job.” These differences among individuals suggest that making sense may serve different functions for different individuals or at least may influence the behavior of some individuals more than others. Although clearly anecdotal, this informally gathered information suggests multiple interesting questions to pursue in future studies.

In an effort to better understand individual differences in sense- making, many future research directions are implicated. An interesting and important next step would be to investigate whether experiential avoidance can distinguish individuals that prefer solvable conditions from individuals who prefer neutral conditions. The Acceptance and Action Questionnaire (AAQ; Hayes et al., 2004) is the most commonly used measure of experien-tial avoidance, has been reliably shown to correlate with clinical problems (Hayes, Luoma, Bond, Masuda, & Lillis, 2006), and poses one promising possibility for measuring the relationship between experiential avoidance and sense- making. For example, individuals who are high in experiential avoidance may be most likely to prefer neutral conditions, which allow the avoidance of unsolvable conditions that likely evoke unpleasant emotional responses without requiring ongoing problem solving.

Second, while the group data presented here collectively provide reliable preference data, the current experimental procedure did not produce reliable individual subject effects, which may be the result of a number of factors. First, the extent to which the dif-ference between the S and U conditions was discernible was not directly assessed apart from participants’ preferences. Although yoking provided the highest level of control across conditions, it may have led some participants to behave as if the U task was solv-able. Specifically, as participants neared the end of the solvable task, most of their responses produced correct feedback, because feedback in the U tasks was yoked to responding in the S task. The high density of reinforcement on latter trials in the U task may have led participants to behave as if both tasks were solvable. Future research should use an unsolvable task that is more discernible from the solvable task. One promising alter-native to evaluate this question would be to use an unsolvable task procedure that punishes any sense- making behavior while reinforcing only random responding. A second alterna-tive would be to deliver reinforcement in the unsolvable task randomly rather than yoking. However, maximizing the ability to discern the unsolvability of the U task might require methodological changes that would sacrifice experimental control. Specifically, to control for the effects of the amount of programmed reinforcement on participants’ preferences, it is critical that the total amount of programmed reinforcement be identical across the two tasks.

Another factor that may have diminished within- subject differences was the wording of the instructions, which may have led participants to believe that both the S and U tasks were solvable. Specifically, the instructions read to the participants encouraged problem

616 WRAY ET AL.

solving in the S and U tasks and discouraged problem solving in the neutral tasks. Thus, the wording of the instructions may have inadvertently contributed to participants’ inabil-ity to discern differences between the S and U tasks. Future research in this area would benefit from additional experimental analysis to identify and isolate the variables respon-sible for participant preference.

Finally, individuals are motivated to assign causality to events that are relevant to them, particularly aversive events (Peterson & Seligman, 1984; Schachter & Singer, 1962). Given that ambiguous or unsolvable conditions are likely experienced as aversive, individu-als may be particularly motivated to assign meaning under these sets of conditions. Participants’ verbal reports indicate that even in the U task, several participants created a sensible narrative about this condition. For example, in the debriefing, participants often spoke about idiosyncratic rules they had created about how the task could give conflicting feedback (e.g., “More than one answer was correct in the U task.”). This finding is similar to the idea that when situations are uncontrollable, individuals will impose control through verbal means (Rothbaum, Weisz, & Snyder, 1982). This is particularly interesting because the meaning imposed was not accurate, nonetheless it may have successfully reduced the uncomfortable emotional state associated with the U task. This may have contributed to any difficulty discerning the S and U tasks as participants were imposing a sensible narrative on an unsolvable situation. The extent to which participants fashioned meaning out of a mean-ingless situation further indicates the pervasiveness of sense- making. However, these hypotheses are only speculative, and future research is needed to evaluate these questions.

In summary, individuals appear to prefer contexts that occasion effective sense- making over those that do not, and this preference seems to be driven in general by the negatively reinforcing effects of avoiding the latter. This may help explain why some individuals engage in sense- making when it is unhelpful and even maladaptive. Although this study is a first step in identifying the consequences that maintain sense- making, future research that more fully assesses this behavioral dynamic would certainly be conceptually useful and may also suggest ways to intervene clinically when sense- making becomes problematic.

ReferencesADDIS, M. E., & CARPENTER, K. M. (1999). Why, why, why? Reason- giving and

rumination as predictors of response to activation- and insight- oriented treatment rationales. Journal of Clinical Psychology, 55, 881–894.

ADDIS, M. E., & JACOBSON, N. S. (1996). Reasons for depression and the process and outcome of cognitive- behavioral psychotherapies. Journal of Consulting and Clinical Psychology, 64, 1417–1424.

AFFLECK, G., TENNEN, H., CROOG, S., & LEVINE, S. (1987). Causal attribution, perceived benefits, and morbidity after a heart attack: An 8-year study. Journal of Consulting and Clinical Psychology, 55, 29–35.

ANTONOVSKY, A. (1979). Health, stress, and coping. San Francisco, CA: Jossey- Bass.AVERILL, J. R. (1973). Personal control over aversive stimuli and its relationship to

stress. Psychological Bulletin, 80, 286–303.BECK, A. T. (1967). Depression: Causes and treatment. Philadelphia: University of

Pennsylvania Press.BECK, A. T., RUSH, A. J., SHAW, B. F., & EMERY, G. (1979). Cognitive therapy of depression.

New York, NY: Guilford Press.BIRD, A., & REESE, E. (2006). Emotional reminiscing and the development of an

autobiographical self. Developmental Psychology, 42, 613–626.BORKOVEC, T. D., ROBINSON, E., PRUZINSKY, T., & DEPREE, J. A. (1983). Preliminary

exploration of worry: Some characteristics and processes. Behaviour Research and Therapy, 21, 9–16.

617MAKING SENSE

BOWER, J. E., KEMENY, E., TAYLOR, S. E., & FAHEY, J. L. (1998). Cognitive processing, discovery of meaning, CD4 decline, and AIDS- related mortality among bereaved HIV- seropositive men. Journal of Consulting and Clinical Psychology, 66, 979–986.

BROCKNER, J., & HULTON, A. (1978). How to reverse the vicious cycle of low self- esteem: The importance of attentional focus. Journal of Experimental Social Psychology, 14, 564–578.

BRUNER, A., & REVUSKY, S. (1961). Collateral behavior in humans. Journal of the Experimental Analysis of Behavior, 4, 349–350.

BURGER, J. M., & ARKIN, R. M. (1980). Prediction, control and learned helplessness. Journal of Personality and Social Psychology, 38, 482–491.

BUSS, A. (1980). Self- consciousness and social anxiety. San Fransisco, CA: W. H. Freeman.

CALHOUN, L., & TEDESCHI, R. (2000). Early posttraumatic interventions: Facilitating possibilities for growth. Posttraumatic stress intervention: Challenges, issues, and perspectives (pp. 135–152). Springfield, IL: Charles C. Thomas.

CARVER, C. S., & SCHEIER, M. F. (1981). Attention and self- regulation: A control- therapy approach to human behavior. Berlin, Germany: Springer- Verlag.

CARVER, C. S., & SCHEIER, M. F. (1986). Functional and dysfunctional responses to anxiety: The interaction between expectancies and self- focused attention. In R. Schwartzer (Ed.), Self- related cognitions in anxiety and motivation (pp. 111–141). Hillsdale, NJ: Erlbaum.

CATANIA, C. A. (1992). Reinforcement schedules. In J. Sanns (Ed.), Learning (3rd ed., pp. 157–185). Englewood Cliffs, NJ: Prentice- Hall.

CHAPMAN, I., & CHAPMAN, J. (1967). Genesis of popular but erroneous psychodiagnostic observations. Journal of Abnormal Psychology, 72, 193–204.

DUVAL, S., & WICKLUND, R. (1972). A theory of objective self awareness. Oxford, England: Academic Press.

ELLIS, A. (1962). Reason and emotion in psychotherapy. Secaucus, NJ: Citadel.ELLIS, A. (1990). Special features of rational- emotive therapy. In W. Dryden & R.

DiGuisseppe (Eds.), A primer on rational- emotive therapy (pp. 79–93). Champaign, IL: Research Press.

GERGEN, K. J., & GERGEN, M. M. (1988). Narrative and the self as relationship. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 21, pp. 17–56). New York, NY: Academic Press.

GERWIRTZ, J. L., & STENGLE, K. G. (1968). Learning of generalized imitation as the basis for identification. Psychological Review, 5, 374–397.

GLASS, D. C., & SINGER, J. E. (1972). Urban noise: Experiments on noise and social stressors. New York, NY: Academic Press.

GOLDING, S. L., & RORER, L. G. (1972). “Illusory correlation” and the learning of clinical judgment. Journal of Abnormal Psychology, 80, 249–260.

HARRISON, R. J., & GREEN, G. (1990). Development of conditional and equivalence relations without differential consequences. Journal of the Experimental Analysis of Behavior, 54, 225–237.

HAYES, S. C., BARNES- HOLMES, D., & ROCHE, B. (2001). Relational frame theory: A post- Skinnerian account of human language and cognition. New York, NY: Kluwer Academic/Plenum Publishers.

HAYES, S. C., LUOMA, J., BOND, F., MASUDA, A., & LILLIS, J. (2006). Acceptance and commitment therapy: Model, processes, and outcomes. Behaviour Research and Therapy, 44, 1–25.

618 WRAY ET AL.

HAYES, S. C., STROSAHL, K. D., & WILSON, K. G. (1999). Acceptance and commitment therapy: An experiential approach to behavior therapy. New York, NY: Guilford Press.

HAYES, S. C., STROSAHL, K. D., WILSON, K. G., BISSETT, R. T., PISTORELLO, J., TOARMINO, D., ... MCCURRY, S. M. (2004). Measuring experiential avoidance: A preliminary test of a working model. The Psychological Record, 54, 553–578.

HAYES, S. C., & WILSON, K. G. (1994). Acceptance and commitment therapy: Altering the verbal support for experiential avoidance. The Behavior Analyst, 17, 289–303.

HOLYOAK, K. J., & SIMON, D. (1999). Bidirectional reasoning in decision making by constraint satisfaction. Journal of Experimental Psychology, 128, 3–31.

HULL, J. (1981). A self- awareness model of the causes and effects of alcohol consumption. Journal of Abnormal Psychology, 90, 586–600.

INGRAM, R. E. (1990). Self- focused attention in clinical disorders: Review and a conceptual model. Psychological Bulletin, 107, 156–176.

JANOFF- BULMAN, R., & FRIEZE, I. (1983). A theoretical perspective for understanding reactions to victimization. Journal of Social Issues, 39, 1–17.

KELLEY, H. H. (1972). Attribution in social interaction. In E. E. Jones, D. E. Kanouse, H. H. Kelley, R. E. Nisbett, S. Valins, & B. Weiner (Eds.), Attribution: Perceiving the cause of behavior (pp. 1–26). Morristown, NJ: General Learning Press.

KELLEY, H. H. (1973). The processes of causal attribution. American Psychologist, 28, 107–128.

LEWINSOHN, P. M., HOBERMAN, H., TERI, L., & HAUTZINGER, M. (1985). An integrative theory of depression. In S. Reiss & R. Bootzin (Eds.), Theoretical issues in behavior therapy (pp. 331–359). New York, NY: Academic Press.

LYUBOMIRSKY, S., & NOLEN- HOEKSEMA, S. (1993). Self- perpetuating properties of dysphoric rumination. Journal of Personality and Social Psychology, 65, 339–349.

MARTIN, L. L., & TESSER, A. (1996). Some ruminative thoughts. In R. S. Wyer, Jr. (Ed.), Ruminative thoughts (pp. 1–47). Mahwah, NJ: Lawrence Erlbaum.

MAXWELL, S. E., & DELANEY, H. D. (2004). Designing experiments and analyzing data: A model comparison perspective (2nd ed.). Mahwah, NJ: Lawrence Erlbaum Associates.

MCADAMS, D. P., REYNOLDS, J., LEWIS, M., PATTEN, A., & BOWMAN, P. J. (2001). When bad things turn good and good things turn bad: Sequences of redemption and contamination in life narrative, and their relation to psychosocial adaptation in midlife adults and in students. Personality and Social Psychology Bulletin, 27, 472–483.

MENDOLA, R., TENNEN, H., AFFLECK, G., MCCANN, L., & FITZGERALD, R. (1990). Appraisal and adaption among women with impaired fertility. Cognitive Therapy and Research, 14, 79–93.

MINEKA, S., & HENDERSON, R. W. (1985). Controllability and predictability in acquired motivation. Annual Review of Psychology, 36, 495–529.

MUSSON, R., & ALLOY, L. (1988). Depression and self- directed attention. Cognitive processes in depression (pp. 193–220). New York, NY: Guilford Press.

NATALE, M., DAHLBERG, C., & JAFFE, J. (1978). The relationship of defensive language behavior in patient monologues to the course of psychoanalysis. Journal of Consulting and Clinical Psychology, 46, 352–353.

NOLEN- HOEKSEMA, S. (1991). Response to depression and their effects on the duration of depressive episode. Journal of Abnormal Psychology, 100, 569–582.

619MAKING SENSE

PALS, J. (2006). Constructing the “springboard effect”: Causal connections, self- making, and growth within the life story. In D. P. McAdams, R. Josselson, & A. Lieblich (Eds.), Identity and story: Creating self in narrative (pp. 175–199). Washington, DC: American Psychological Association.

PAPAGEORGIOU, C., & SIEGLE, G. (2003). Rumination and depression: Advances in theory and research. Cognitive Therapy and Research, 27, 243–245.

PAPAGEORGIOU, C., & WELLS, A. (2001). Metacognitive beliefs about rumination in recurrent major depression. Cognitive and Behavioral Practice, 8, 160–164.

PAPAGEORGIOU, C., & WELLS, A. (2003). An empirical test of a clinical metacognitive model of rumination and depression. Cognitive Therapy and Research, 27, 261–273.

PARK, R. J., GOODYER, I. M., & TEASDALE, J. D. (2004). Effects of rumination and distraction on mood and overgeneral autobiographical memory in adolescent major depressive disorder and controls. Journal of Child Psychology and Psychiatry, 45, 996–1006.

PENNEBAKER, J. W., & SEAGAL, J. D. (1999). Forming a story: The health benefits of narrative. Journal of Clinical Psychology, 55, 1243–1254.

PETERSON, C., & SELIGMAN, M. E. P. (1984). Causal explanations as a risk factor for depression. Psychological Review, 91, 347–374.

PYSZCZYNSKI, T., & GREENBERG, J. (1986). Evidence for a depressive self- focusing style. Journal of Research in Personality, 20, 95–106.

PYSZCZYNSKI, T., & GREENBERG, J. (1987). Self- regulatory perseveration and the depressive self- focusing style: A self- awareness theory of reactive depression. Psychological Bulletin, 102, 122–138.

ROTHBAUM, F., WEISZ, J., & SNYDER, S. (1982). Changing the world and changing the self: A two- process model of perceived control. Journal of Personality and Social Psychology, 42, 5–37.

SARASON, I. G. (1975). Anxiety and self- preoocupation. In I. G. Saranson & C. D. Spielberger (Eds.), Stress and anxiety (pp. 129–151). New York, NY: Hemisphere.

SARASON, I. G. (1986). Test anxiety, worry, and cognitive interference. In R. Schwarzer (Ed.), Self- related cognitions in anxiety and motivation (pp. 19–33). Hillsdale, NJ: Erlbaum.

SCHACHTER, S., & SINGER, J. (1962). Cognitive, social, and physiological determinants of emotional state. Psychological Review, 69, 379–399.

SKINNER, B. F. (1936). The verbal summator and a method for the study of latent speech. Journal of Psychology, 2, 71–107.

SKINNER, B. F. (1953). Science and human behavior. New York, NY: The Macmillan Company.

SOSNOWSKI, T. (1983). Sources of uncertainty and psychophysiological state: Changes under conditions of noxious stimulus anticipation. Biological Psychology, 17, 297–309.

SOSNOWSKI, T. (1988). Patterns of skin conductance and heart rate changes under anticipatory stress conditions. Journal of Psychophysiology, 2, 231–238.

STARR, B. J., & KATLIN, E. S. (1969). The clinical as an aberrant actuary: Illusory correlation and incomplete sentences blank. Journal of Abnormal Psychology, 74, 670–675.

SUNDBERG, M., & MICHAEL, J. (2001). The benefits of Skinner’s analysis of verbal behavior for children with autism. Behavior Modification, 25, 698–723.

VAUGHAN, M. E., & MICHAEL, J. (1982). Automatic reinforcement: An important but ignored concept. Behaviorism, 10, 217–227.

620 WRAY ET AL.

WARD, W. C., & JENKINS, H. M. (1965). The display of information and the judgment of contingency. Journal of Psychology/Revue Canadienne de Psychologie, 19, 231–241.

WATKINS, E. (2008). Constructive and unconstructive repetitive thought. Psychological Bulletin, 134, 163–206.

WINE, J. D. (1971). Test anxiety and direction of attention. Psychological Bulletin, 76, 92–104.

WINE, J. D. (1982). Evaluation anxiety: A cognitive- attentional construct. In H. W. Krohne & L. Laux (Eds.), Achievement, stress, and anxiety (pp. 207–219). Washington, DC: Hemisphere.

WRAY, A. M., & DOUGHER, M. J. (2009, May). Reinforcing functions of making sense using socially relevant stimuli. Poster session presented at the annual meeting of the Association for Behavior Analysis International Convention, Phoenix, AZ.

WRAY, A. M., DOUGHER, M. J., & BULLARD, L. (2008, May). Examining the reinforcing effects of making sense. Poster session presented at the annual meeting of the Association for Behavior Analysis International Convention, Chicago, IL.

WRAY, A. M., FREUND, R. A., & DOUGHER, M. J. (2009). A behavior analytic account of cognitive bias in clinical populations. The Behavior Analyst, 31, 29–49.

621MAKING SENSE

Appendix A

Individual Participant Preference Data for Experiment 2

ID # # Choice S # Choice N # Choice U

Participant 2-1 2 4 2


Participant 2-3 2 4.5 1













Note. S = solvable; N = average of N1 and N2; U = unsolvable.

622 WRAY ET AL.

Appendix B

Individual Participant Preference Data for Experiment 3

ID # # Choice S # Choice N # Choice U





















Note. S = solvable; N = average of N1 and N2; U = unsolvable.

Examining the Reinforcing Properties of Making Sense: A Preliminary Investigation

Documents