reared in a restricted environment. Journal of Comparative &Physiological Ps ychology, 1970,70,157-164 .

Montgomer y, K ., & Zimbardo, · P . Effect of sensory andbehavioral d epriva ti on upon ex ploratory behavior in the rat.Perceptual & Mo tor Skills, 1957 , 7 , 223-229.

Neuri n ger , A . J . Animals respond for food in the presence of fr eefo od. S ci ence, 1969, 166. 399-401.

R os en zweig, M. R oo Bennett, E . L ., & Diamond, M. C . Brainchanges in response t o experience . Scientific American, 1972,226,22-29.

Tarte, R. D., & Snyder, R . L. Barpressing in the presence of fr eefo od as a function of food d eprivation. Psvch onomlc Science,1972,26,169-170.

Walker, E . L . Ps ychological complexity as a b asis for a theory o f

m otivation and choice. In D . Levine (Ed .) , N ebrask asy m posiu m o n motivation. Lincoln, Nebr: University o fNebrask a Press , 1964. Pp, 47-96.

Wo ods, P . J., Fiske, A . S. , & Ruck elhaus, S . I. The effects ofdrives c on fl ic ti n g with exploration on the problem-solvingbehavi or o f rats reared in fr ee and restricted environments.Journal of C omparative & Physiological Ps ychology , 1961 ,54,167-169.

Zimbardo, P. G ., & Montgomery, K . C . Effects o f " free"e nv i ro n m e n t a l rear ing upon exploratory behavior.Ps ychol ogical Reports, 1957, 3, 589-594.

(Received for publication May 10, 1973.)

The effects of discrimination training on the recognition ofwhite and oriental faces

ELAINE S. ELLIOTT, ELIZABETH J. WILLSand

ALVING. GOLDSTEINUniversity ofMissouri, Columbia, Mo. 6520 1

Recognition for faces of whites and orientals wasinvestigated using white Ss at a predominantly wh iteuniversity. Ss who participated in a discriminationtraining, or practice, session involving a paired-associate(PA) task with oriental faces scored significantly higheron a subsequent recognition test involving oriental facesthan did Ss who received no practice or Ss who received

Bull. Psychon. Soc., 1973, Vol. 2 (2)

practice involving white faces . Ss who practiced withwhite faces did not score significantly higher on therecogni tion test than Ss without practice or Ss who saworiental faces during the practice session. These resultslend support to a differential experience hypothesis,with the possibility of schema functioning as a mediatorof differential experience.

Maplass & Kravitz (1969) attempted to demonstratedifferential recognition performance for faces of ownand other race among both white and black Ss from a

71

predominantly black and predominantly whiteuniversity. They found that white faces were recognizedmore often than were black faces, but that there was asignificant interaction between race of S and race of thestimulus . These findings are consistent with a hypothesisof differential experience with persons of other race anddifferential acquisition of cue utilization habits.However, the questionnaire data failed to show acorrelation between reported experience and recognitionperformance. Alternatively, these authors suggested thatthe source of the difficulty in the recognition of theblack face by white Ss lies in the darkness of the blackface, which obliterates many discriminative details.However, in a study testing the recognition performanceof black and white Ss for white, black, and orientalfaces, Chance, Goldstein, and Mcllrlde! found thatwhites recognized white faces best and oriental facesleast, whereas blacks recognized black faces best andoriental faces least. The fact that oriental faces were lesswell recognized than black faces argues against the ideathat poor recognition of black faces by whites lies in thedarkness of the skin . These results are compatible with adifferential experience hypothesis, since blacks andwhites have had more experience with their own racesand much less with the oriental race (the experiment wasconducted at a Midwestern university) .

Although the studies cited suggest a differentialexperience hypothesis as a possible explanation ofdifferential recognition for faces of ethnic groups , theydid not test this hypothesis directly by providingexperimentally defined differential experience indiscrimination. Results of other perceptual studies (e.g.,Gibson , 1953) indicate that discrimination andrecognition performance in general can be affected byprior experience through training.

The present study attempts to test the hypothesis thatdifferential experience in making a discriminationbetween faces of a race different from S's will affectlater recognition performance for faces of that race. Adiscrimination training, or practice, session will beconducted during which Ss will have to discriminateamong a group of white faces or a group of oriental facesin a picture-number PA task . Assuming that white ,midwestern college students have had little experiencewith orientals, the differential experience hypothesis willgain support if Ss given practice in discriminatingbetween oriental faces score higher on a subsequentrecognition task involving oriental faces than Ss whoreceive no training . Since these Ss will have had a greatdeal of experience in discriminating among white faces,their recognition performance for white faces should notbe improved after discrimination training involving whitefaces.

METHODSubjects and Design

Sixty-six white students were recruited from an introductorypsychology class to serve as Ss in partial fulfillment of courserequirements. Ss were assigned randomly to one of six groups.

72

Twelve Ss participated in each of four discriminat ion training. orpractice. groups. Two groups were involved in a discriminationtask using white faces and were then tested on either white ororiental faces in a recognition task, and two groups were exposedto a discrimination task with oriental faces and were then given arecognition task involving either oriental or white faces. Twogroups of 8 and 10 Ss each were simply tested on a recognitiontask using white or oriental faces, respectively, to determinebase-line recognition performance for the stimuli used in theexperiment.

Materials and ApparatusStimuli used to determine base-line recognition and for testing

recognition after practice were 80 black-and-white Caucasoid ororiental projected images. Half of the stimuli were female, andhalf were male. All photos showed a full-face view of thestimulus. A Carousel projector was used to present the slides.

An additional 30 Caucasian and 30 oriental faces mounted on4 x 6 in cards were used in the PA task. In both sets of stimuli,each picture was mounted on one side of a card and a two-digitnumber 00 through 39) was paired with it on the other side ofthe card. The photos of white adults were obtained from aschool district in the U.S., whereas the pictures of oriental adultswere obtained from a newspaper article on China.

ProcedurePA Task

All Ss who were given practice via the PA task were testedindividually. Each group of 30 white and 30 oriental stimuli wasarbitrarily divided into three sets of 10 faces. Twenty-four Sslearned oriental picture-number PAs, whereas 24 learned whitepicture-number PAs. Ss were told that they would be given 10cards at a time and that their task was to learn which face waspaired with each number. Each S was given approximately 2 minto study each set of PAs. E then showed each picture to S forapproximately 4 sec, during which time S had to respond orallywith the number paired with the picture. If 5 failed to respondor responded incorrectly, E said the correct number .

This procedure continued for either 10 trials or to a criterionof 2 consecutive errorless trials. The procedure was repeated forthe remaining two sets of 10 cards. The oriental base-line andwhite base-line groups did not participate in the PA task.

Recognition TestBoth oriental base-line Ss and white base-line Ss were tested in

groups, whereas Ss who received practice were testedindividually immediately following the PA task. Ss were showneither 20 Caucasian or 20 oriental faces at a rate of 3-4 sec perslide. Ss were instructed to attend to the slides and nothing else.Recognition memory was measured immediately for all Ss andthe method for testing recognition was the same for all groups.

In the recognition test , each S was shown a total of 80 stimulifor 3-4 sec each. Sixty of the slides were "new" (not seenbefore) and 20 were "old" (seen before) . Presentation of old andnew stimuli was random and identical for all Ss, Responses wererecorded by S on an answer sheet numbered 1·80, with twocolumns headed "Seen Before" and "Not Seen Before." S wasnever informed of his accuracy .

The problem of response bias was handled by explicitlyinstructing S that of the 80 photos, exactly 20 would be"seen-before" items. Ss were asked to try to restrict the numberof their seen-before responses to 20. Compliance with theseinstructions was facilitated by the single-page open format of theanswer sheet , clearly numbered for 80 responses, whichpermitted S to readily count his seen-before responses.

RESULTS AND DISCUSSIONIn a recognition study, false alarms are not of critical

importance as long as the Ss comply with instructions to

Bull . Psychon. Soc ., 1973, Vol. 2 (2)

(Received for publication May 16, 1973.)

NOTE1. Chance, J .• Goldstein. A. G., & McBride, L . Faces and

races: A series of explorations. Unpublished paper.

Table 1Means and Standard Deviations of the Number of

Correct Recognition Responses

REFERENCESAttneave, F. Transfer of experience with a class-schema to

identification-learning of patterns and shapes. Journal ofExperimental Psychology, 1957,54,81-88.

Gibson, E . J . Improvement in perceptual judgments as a functionof controlled practice or training. Psychological Bulletin,1953,50, 401-431.

Malpass, R. S., & Kravitz, J . Recognition for faces of own andother race. Journal of Personality & Social Psychology, 1969,13, 330-334.

Vernon, M. D. The functions of schemata in perceiving.Psychological Review , 1955. 62. 180-192.

2.4122.5032.4292.5592.3532.587

SDMean

10.4011.4113.5815.3715.5815.16

Group

No Practice/Oriental RecognitionWhite Practice/Oriental RecognitionOriental Practice/Oriental RecognitionNo Practice/White RecognitionOriental Practice/White RecognitionWhite Practice/White Recognition

Note- The maximum number of correct recognition responsespossible is 20.

The above findings are compatible with a differentialexperience hypothesis. It appears that thediscrimination, which was of instrumental value in theoriental PA task , was sufficient to bring recognition fororiental faces equal to the recognition level for whitefaces. Perhaps this group developed a schema for theoriental face during practice. Through discrimination inthe PA training task, Ss acquired a greater knowledge ofthe class of oriental faces. It is probable that throughoccasional exposure to orientals, Ss had a partial schemaof the class of oriental faces, a notion of its centraltendency, which allowed them to make between-groupdiscriminations. However, as this schema did not satisfya need, it remained suppressed (Vernon, 1955), whileschema for white faces, which were instrumental inday-to-day experience, were built up. Throughdiscrimination training with oriental faces, Ss acquired aknowledge of the schema's remaining characteristics(Attneave, 1957). They learned how individual orientalfaces may differ (the properties and dimensions ofvariability) and how much individual oriental faces varyfrom the average or central tendency. Perhaps these arethe parts of schema which allow for within-groupdiscriminations.

In summary, this study of differential recognitionperformance for faces has indicated that recognition fororiental faces may be improved with practice indiscriminating among oriental faces. These findings lendsupport to a differential experience hypothesis withschema functioning as an entity to mediate differentialexperience.

limit their "seen-before" responses to 20. The meannumber of old responses, regardless of correctness, was20 in the present study. This mean is very representative,since nearly all Ss restricted themselves to 18, 19, or 20responses. Therefore, recognition scores of correctresponses should be an adequate estimate of recognitionhere.

Means and standard deviations of the number ofcorrect recognition responses made by each of the sixexperimental groups are presented in Table 1.

As expected, analysis of data showed that correctrecognition performance of the group that viewed whitefaces without practice (76.8%) was significantly betterthan the recognition performance of the group thatviewed oriental faces without practice [52%;t(16) = 4.23, p < .01]. These findings are in accord withother studies which have found that recognitionperformance for white faces is better than recognitionperformance for oriental faces by white Ss (Chance,Goldstein, & McBride1 ) . The oriental group withpractice in discriminating among oriental facesperformed significantly better on the orientalrecognition task (68%) than did the group that wastested for recognition of oriental faces after practicewith white faces [57%; t(22) = 2.15 , p < .05] and thegroup that was tested on oriental faces without practice[52%; t(20) =3.07, P < .01]. This finding supports ahypothesis of differential experience in that S performedsignificantly better after gaining some experience indiscriminating among oriental faces. Since the group thatviewed white faces during practice and oriental facesduring the recognition task did not perform significantlybetter than the group that viewed oriental faces duringthe recognition task without practice [t(20) = .964,p > .05] , one can reject the idea that training enabled Sto develop a set for learning, i.e., learning-to-learn orwarm-up.

Further support for the rejection of warming up as apossible explanation for the improvement of the groupthat viewed oriental faces during practice and duringrecognition lies in the fact that significant performancedifferences were not found between the group thatviewed white faces during recognition but who did notparticipate in the PA task (76 .8%) and the group thatviewed white faces during both practice and recognition[76%; t(18) = .177 , p > .05] or between the group thatviewed white faces during recognition without practice(76.8%) and the group that viewed oriental faces duringpractice and white faces during recognition [78%;t(18) = .187, P > .05] . Furthermore, a significantdifference was not found between the groups thatviewed oriental faces during both the PA test and therecognit ion test and the group that viewed white facesduring both tests [t(22) = 1.54 , p > .05] . Positivetransfer from the oriental PA test to the orientalrecognition test and the lack of positive transfer fromthe white PA test to the white recognition test resultedin no significant difference between the two groups.

Bull. Psychon. Soc., 1973, Vol. 2 (2) 73