Example 4 unannotated)ibpsychologymrpoll.weebly.com/uploads/3/0/3/4/... · faces and sketches of costumes, which found that inversionof faces had a greater effect on recognition.

Psychology teacher support material

Example 4 (unannotated)

1

Face recognition

1982 words

Introduction

Cognitive psychology is the study of processes involved in making sense of the environment and interacting appropriately with it. Memory, the cognitive process by which we retain and recall information, and perception, the process by which we receive and make sense of information from our environment, are processes both involvedin face recognition. Many studies investigate the involvement of these processes on face recognition.

A widely accepted theory that has arisen from studies on face recognition is the holistic form theory, which suggests that we look at faces as a whole to recognize them. This involves a wider range of cognitive processes, a much more complex process than simple feature analysis, which says that we recognize faces by individual features.

Many studies have shown that human faces are more difficult to recognise when inverted, for example a study byKohler (1940, cited in Yin, 1969), who thought this was due to the "loss of facial expression" when the picture was shown upside-down. This was supported by further research (Goldstein & Brooks 1963, cited in Yin 1969) in which children found it more difficult to recognize inverted photos of classmates than upright photos.

Studies by Yin (1969) investigated recognition of objects which are usually only seen upright, such as houses and airplanes, in addition to faces. Upright and inverted photographs of objects and faces were presented to participants, and participant's memories of these were tested and compared. Yin found that both the objects and faces were easier to remember when upright, but that the effect of inversion was significantly greater in faces than it was inobjects. This was reinforced by Yin's later experiment (1969) comparing the effects of recognition in drawings of faces and sketches of costumes, which found that inversion of faces had a greater effect on recognition. This experiment also showed that the effects of inversion were not limited to photographs but also existed in drawings,which supplements evidence indicating the difficulty of recognising upside-down faces.

Yin's experiments (1969) support the interpretation that this difficulty is linked to a general difficulty inrecognising objects which are orientated unfamiliarly, and also a different determinant linked to faces. This implies that the cognitive processing underlying recognition of faces differs to that of object recognition.

This experiment aims to investigate the effects of inversion on face recognition.

A one tailed directional hypothesis was used in this experiment as the theory on which it is based is considered robust.

The Independent Variable was the different orientation of faces in the conditions; upright and inverted faces. The Dependent Variable was the time taken in seconds for the participants to recognise the faces.

Experimental hypothesis There will be a significant increase in time taken, in seconds, when participants are asked to recognise an inverted face compared to a non-inverted face.



2

Null hypothesis There will be no significant increase in time taken, in seconds; when participants are asked to recognize an inverted face compared to a non-inverted face. Any differences are down to chance factors alone.



3

Exploration

The experimental design used was independent samples. The experiment was carried out in the form of a field experiment for high ecological validity because it was conducted in the participants’ natural environment. It also has reduced demand characteristics because the aim of the study is not obvious.However there-may havebeen lower levels of control due tovarious extraneous variables suchasbackgroundnoiseandthetimeofday oftheexperiment which could potentially affect participant's concentration andperformance (discussed further in 'Limitations and Modifications'). It is therefore more difficult to obtain accurate results than with a laboratory experiment.

In terms of ethical issues an element of deception was necessary to prevent demand characteristics but the participants gave partially informed consent by signing a consent form after reading instructions, and were advised of their right to withdraw. They gave fully informed consent and consent to the data being used. There was no harm caused by the study and all participants were informed that their data was in the normal range at the debrief. No identifying details were kept on the participants for the sake of confidentiality.

The elements of control were within the two pages displaying 10 photographs of famous faces for each condition. All photographs were identically displayed for the two conditions, the only difference being the inversions. To ensure the celebrities were well known, the photos were initially tested for recognisability in a pilot test. The pages could not be moved by participants, as a change in angle would alter the orientation of the photos and could affect the difficulty of recognising the faces, so this needed to be controlled.

The target population were Further Education students aged 16 – 19. A total of 21 participants were used, 10 male and 11 females. Opportunity sampling was used as this is the most convenient method to use in the limited time available. Random representative sampling was considered but not used due to the time issues involved. Participants were randomly allocated to a condition to ensure that the participants had fairly similar characteristics between the two groups, because there was an equal chance that participants were selected for either condition. Psychology students were not asked in case they guess the nature of the study and changed their behaviour.

Materials

• 2 pages showing 10 faces, x2: either upright or inverted • Stopwatch• One-penny coin to assign conditions • Brief • Standardised instructions

• Consent forms• Accepted answers list • Recording sheet

• Debrief

Procedure

The experiment was carried out in a library, from 10.30-l 1.30am and 2.45-4.15pm and conducted as follows:



4

i. Random assignment of participant's condition by the flip of a coinii. Participants were given standardised instructions and briefediii. On giving consent, participants were presented with 2 blank sheets of paper. Sheets were placed next to

each other with no space between them. Underneath, another 2 sheets displayed the inverted or upright faces, depending on the assigned condition.

iv. iv. The blank sheets covering the faces were simultaneously removed and the stopwatch was started.v. In sequence (110), pa1ticipants attempted to name all faces. Answers were recorded on the recording sheet.

Incorrect answers, i.e. those not corresponding with a name on the accepted answers list, or failure to provide an answer, included a 2 second penalty, which was added on to participant's time score at the end.

vi. The timer was stopped after the participant gave their final answer.vii. Participants were debriefed.

The blank sheets covering the faces were simultaneously removed and the stopwatch was started.In sequence (110), pa1ticipants attempted to name all faces. Answers were recorded on the recording sheet (see appendix VII). Incorrect answers, i.e. those not corresponding with a name on the accepted answers list, or failure to provide an answer, incured a 2 second penalty, which was added on to participant's time score at the end. Analysis

Descriptive Statistics

The appropriate measure of central tendency was the mean and it suggests the calculation of standarddeviation to measure the level of dispersion.

(See appendix IX for raw data.)

Table 1: to show the mean time taken, in seconds to name the faces in the two conditions; upright faces condition and inverted faces condition ,

Upright Faces Condition Inverted Faces Condition

Mean Time Taken (seconds) 11.67 24.39Standard Deviation 3.32 5.69

(See appendix X for mean and standard deviation calculations.)

Table 1 shows that, on average, participants took 12.72 seconds longer to recognize inverted famous facescompared to the non-inverted faces. It shows a substantial difference between the conditions and greater dispersion when faces were inverted.

The non-inverted faces condition was more evenly spread than the inverted faces condition,in which thespread of data was slightly larger, showing more variance in participant's ability to recognise inverted faces.



5

Graph 1 and Table 1 show that, on average, it clearly takes longer to identify inverted facescompared to upright faces so suggests the experimental hypothesis is supported but inferentialstatistics must be completed to see if the difference between the conditions is significant.

Inferential Statistics

The test that should have been used to compare two populations of data was the unrelated T test, but as the data was not normally distributed Mann Whitney U test was used.Table 2: Treatment of Results

Coolican (1999) states that "for any n' and n2, the observed value of U is significant at a given level ofsignificance if it is equal to or less than the critical values shown."The calculated value of U is 2 and the critical value is 31, therefore the value of U is significantly less than

Calculated U value 2

Critical U value 31

Tails 1

P 5%



6

the critical value. This means that the experimental hypothesis that "there isasignificant increase in timewhenparticipants are asked torecognize an inverted face when compared to a non-inverted face", can be accepted and the null hypothesis can be rejected. This infers that there is a significant increase in time when participants are asked to recognize faces that are inverted compared toupright.

The mean length of time taken for non inverted faces condition was 11.67 seconds and for the inverted condition 24.39 seconds. On average the time it took for participants to recognize inverted faces was 12.72 seconds longer, over double, than the upright faces. The standard deviation show that both face orientation conditions were fairly constant and the graph shows the clear difference between them. The inferential statistics suggest a highly significant difference between the 2 conditions.

Evaluation

The results obtained support the findings of the experiment by Yin (1969), supporting thetheory that it is more difficult to recognise inverted faces. It also matches findings by Goldstein &

Brooks (1963, cited in Yin 1969). Although this investigated recognition of children rather than teenagers, the results show consistency with this theory. The results also show how inversionworsens recognition, supportingYin's study (1969) that found worsened recognition in both faces and objects that were inverted. Therefore there is significantevidence suggesting inversion worsens recognition of faces.

A limitation to the research was the extraneous variable of background noise. Research took place in a busy student environment of a library and the noise may have affected the concentration ofparticipants. However having background noise also meant that participants were in their naturalenvironment; the results may have been affected if participants were in a silent area by making them nervous. To control this, the experiment could have taken place in a quieter area within the natural college environment.

Another extraneous variable was the time of day in which the experiment was carried out; this mayhave affected the amount of light in the room or how tired the participants were; either of these factors could affect their performance. This could be modified with a longertime scale for the experiment, so it could be conducted around the same time of day. A time scale would alsoallow for more data collection, increasing validity further.

Some participants did not recognize certain faces or gave the wrong answer, in which case2 seconds was chosen as a suitable penalty to be added to their time score. However the resultsdo not showwhich participants failed to recognize faces, thus the data does not show a definite amountof time taken for participants to carry out the experiment, decreasing validity.

. .



7

Another limitation affecting validity was the possible time advantage whereparticipants could give an accepted first name rather than a full name; however thiswas balanced by the equal chance of this happening in either condition. Also, it was made clear in the brief that full names should be given where possible tocontrol this.

Participants only took part in one condition which slightly decreases the reliabilitybecause one group may have, by chance, been better at recognising faces. However this was controlled though random allocation to conditions, and it was also necessary for the participants to only take part in one condition becauseparticipants would be abletoremember the celebrities if they were to repeat the experiment, as the same faces were usedfor each condition.

The findings of the experiment were similar to Yin's (1969) suggesting that the experimentalhypothesis can be accepted; concluding that it is more difficult to recognise inverted facesthan upright faces.

References

Coolican, H (1994). Research methods and statistics in Psychology. London; Boder Stoughton.

Yin, R. K. (1969). Journal of Experimental Psychology. Vol. 81, No 1, 1-14



8

Appendix 1- Upright famous faces 1

[NOTE: THE IMAGES USED IN THIS APPENDIX HAVE BEEN REMOVED FOR COPYRIGHT

COMPLIANCE]



9



10



11



12



13



14



15



16



17



18

Example 4 unannotated)ibpsychologymrpoll.weebly.com/uploads/3/0/3/4/... · faces and sketches of costumes, which found that inversionof faces had a greater effect on recognition.

Documents