BRAIN AND COGNITION Electrodermal 240-2.52 (1988) 8, Discrimination of Familiar but Not Unfamiliar Faces in Prosopagnosia RUSSELL M. BAUER Department of Clinical Psychology, University of Florida AND MIEKE VERFAELLIE Department of Neurology, University of Florida It has been previously shown that prosopagnosics can electrodermally “recognize” faces they cannot verbally identify and with which they feel no familiarity. This study extended previous results by showing that electrodermal discrimination of faces exists only on a famous face identification task, and not on a matchingto-sample task involving unfamiliar faces. This suggests that electrodermal recognition reflects the activation of stored identity-specific information built up on the basis of past contact with faces, and provides a psychophysiological distinction between familiar and unfamiliar face processing. Implications for cognitive models of face recognition, and for understanding the nature of prosopagnosia, are discussed. 0 1988 Academic Press. Inc. Prosopagnosia is a rare neurobehavioral syndrome in which a patient with brain damage becomes unable to recognize previously familiar persons by visual reference to their facial features (Bodamer, 1947; Bauer & Rubens, 1985). Prosopagnosics recognize faces as faces, but cannot determine specific facial identity. The defect is limited to specific aspects of vision, since such patients achieve immediate recognition when they hear the person’s voice or through attention to extrafacial cues. Because these patients can generally match faces they cannot identify, This work was supported by Grant AA-06203 to the University of Florida and was presented in part at the 14th Annual Meeting of the International Neuropsychological Society, Denver, CO, February 4-8, 1986. Address correspondence and reprint requests to Dr. Russell M. Batter, Department of Clinical Psychology, University of Florida, Box J-165 JHMHC, Gainesville, FL 32610. Dr. Verfaellie is now at the Psychology Service, Veterans Administration Medical Center, 150 S. Huntington Ave., Boston, MA 02130. 0278-2626/88 $3.00 Copyright All rights 0 1988 by Academic Press. Inc. of reproduction in any form reserved. 240 EDR DISCRIMINATION 241 and because they have obvious impairments in learning new facial identities, the idea has emerged that prosopagnosia involves some kind of memory access impairment in which there are defects in both the activation of stored information about faces (Batter, 1984; Bauer & Trobe, 1984; Damasio, Damasio, & Van Hoesen, 1982) and impairments in the formation of new facial memories. Behavioral variations among reported cases of prosopagnosia suggest that such failure could result from (1) subtly altered perception, the products of which do not correspond with available memory representations; (2) destruction of the memory representation itself; or (3) disconnection of anatomic structures subserving facial perception from the structures in which memory representations are stored (cf. Damasio et al., 1982). Neuropsychological investigations of prosopagnosic patients continue to be influenced by recent experimental work concerned with articulating the processes underlying face perception and identification (cf. Bruce & Young, 1986). One area of investigation involves the distinction between familiar vs. unfamiliar face processing (Benton, 1980; Ellis, Shepherd, & Davies, 1979; Hay & Young, 1986; Malone, Morris, Kay, & Levin, 1982; Young, Hay, & Ellis, 1986; Warrington &James, 1967). Prrcepptual analysis of faces involves the parallel extraction of featural and configural information (Sergent, 1986). The targets and results of this perceptual analysis are different for different processing task. For example, there is evidence that internal features are especially important for familiar face recognition, while internal and external features are equally important when processing unfamiliar faces (Ellis et al., 1979). Zdrntijication of a previously familiar face requires additional postperceptual steps in which a memory representation of the face is activated because it matches the incoming percept in some significant way. The exact nature of the information stored in memory is poorly understood. Bruce and Young (1986) suggest that previous experience with a person leads to the formation of an “identity-specific semantic code” which contains information about the individual’s appearance, demeanor, occupation, family background, hobbies, personality, etc. The multimodal information contained in this code can be accessed by visually apprehending the person’s face, by hearing the person’s voice, or by hearing information about him. From a cognitive perspective, the familiar-unfamiliar face distinction relates to presence or absence of stored identity-specific information about viewed faces. Such cognitive models illustrate the complexity of the face recognition process and raise the important question of the level at which the recognition defect arises in prosopagnosia. For example, do prosopagnosics suffer a loss of identity-specific information, or do they have trouble accessing such information from vision? Until recently, evidence bearing on this question existed solely in the form of overt behavioral performance on 242 BAUER AND VERFAELLIE perceptual tests or by reference to verbal identification of previously familiar or famous faces. However, two recent studies have explored psychophysiological responses of prosopagnosics during facial identification and facial perception tasks (Bauer, 1984; Tranel & Damasio, 1985). In these studies, prosopagnosics display greater psychophysiological activation to familiar but unidentified faces than they do to unfamiliar foils. These data suggest that such patients have knowledge about facial identity which is not reflected in verbal report, and thus favor the notion that a memory access impairment is involved. These studies both used facial stimuli that were familiar to the patients prior to illness onset. What would occur if patients were presented with previously unfamiliar faces? In this paper, we present data showing that differential electrodermal responses in prosopagnosia occur on a task requiring the identification of famous personalities, but not on a matchingto-sample recognition task involving unfamiliar faces. On the basis of these results, we argue that electrodermal discrimination of facial identity in prosopagnosia depends upon the prior existence of stored identityspecific information which is capable of matching an incoming face percept. This provides the first psychophysiological evidence for the familiar vs. unfamiliar face distinction. CASE REPORT The patient is a 59-year-old divorced supply technician who was in her usual state of good health until 2 weeks prior to her clinic visit when she suffered sudden severe headache, diplopia, and acute loss of vision on the left. She was admitted to a local hospital where CT scan (Fig. 1) revealed bilateral occipitotemporal infarcts and an angiogram showed 50% stenosis of the left vertebral artery. Neurological examination revealed normal motor, sensory, and cranial nerve functions with no cerebellar signs or pathological reflexes. Mental status evaluation was normal except that she remembered only l/3 items at 5 mitt, and appeared to have poor color discrimination, particularly in the upper quadrants. She was unable to recognize the consulting physicians upon their return for follow-up evaluation, nor could she learn the faces of hospital staff who attended her daily throughout her stay. Ophthalmological evaluation revealed corrected visual acuities of 20/40 OS and 20/70 OD. Pupils were both 3.5 mm with a 3 + light reaction. Confrontation visual fields revealed left homonymous hemianopia, and formal perimetry revealed additional partial right upper quadrantanopia. Neuropsychological exam revealed a dextral, high-school-educated female with a Verbal IQ of 103 on the Wechsler Adult Intelligence Scale and a Memory Quotient of 118 on the Wechsler Memory Scale (WMS). Detailed memory testing revealed a visual recent memory defect with relatively spared auditory-verbal memory. For example, she recalled EDR DISCRIMINATION 243 FIG. 1. CT scan of patient taken on admission. Note bilateral infarctions in the posterior cerebral artery territory. The lesion on the right is larger than the left. 5, 9, 7, 10, and 10 words on five trials of the California Verbal Learning Test (Delis, Kramer, Kaplan, & Ober, 1983). Cued recall was 12/16 both immediately and after a 20-min delay. Delayed recognition was 15/16. On WMS Logical Memory, she was able to recall an average of lo/24 items, normal for her age. In contrast, visually presented material (ReyOsterrieth Complex Figure, WMS Visual Reproduction, Milner Facial Recognition Test) was poorly remembered immediately and after delays of 20 and 90 min. She named real objects with hesitation, but was nearly totally incapable of naming line drawings or photographs. Mild spelling alexia was present. She could not discern action in complex pictures, describing only a portion at a time. She had no gaze paralysis or misreaching 244 BAUER AND VERFAELLIE for visual targets. She could not name colors, appropriately color line drawings, or sort colors across or within hues (Munsell-Farnsworth 100 Hue Test). She was, however, fully capable of performing “verbalverbal” color tasks (e.g., listing color names, naming the appropriate color for common objects). She was profoundly prosopagnosic, and could not identify any of 48 famous faces shown to her (Albert, Butters, & Levin, 1979, modified to mask nonfacial cues). Family pictures were unavailable. Despite failure to learn any of the faces of doctors, nurses, or neuropsychology staff, if spoken to she could immediately recognize even those she had only briefly met. She could not reliably make man vs. woman or black vs. white discriminations, but could always recognize a face as a face. Her descriptions of her percepts indicated fragmented vision with serial scanning of details. She performed all visual tasks, not just those including faces, extremely slowly. On a facial matching-to-sample task (Test of Facial Recognition; Benton, Hamsher, Varney, & Spreen, 1983), she was 80% accurate on trials in which the comparison face was an exact replica of the target (isomorphic trials), but only 20% accurate when the comparison face was presented under different lighting conditions or angles of view than the target (nonisomorphic). Findings suggested with an apperceptive defect with some of the characteristics of “simultanagnosia” (Wolpert , 1924; Luria, 1959). METHOD AND RESULTS Fumous face identi$cation test (FFIT). Sixteen facial slides from the Boston Remote Memory Battery (Albert et al., 1979, with nonfacial cues masked) were individually displayed and five multiple-choice names, only one of which was correct, were verbally presented with each. The slides depicted famous world leaders. movie stars, and TV personalities chosen from categories in which the patient professed interest and knowledge. This was done to ensure that unfamiliarity with the faces or personalities involved could not be the basis of nonrecognition. Alternative names within each item were selected from the same semantic category as the target name. For example, for Betty Gable. the alternatives were Vivan Leigh, June Haver, Maureen O’Hara, and Jane Powell; for Liza Minelli, the alternatives were Helen Reddy, Shirley MacLaine, Barbra Streisand, and Judy Garland. The entire procedure was controlled by an IBM PC/XT microcomputer and a Data Translation DT2805 Analog-to-Digital converter. While each slide was in view, the following events occurred. First, the patient was allowed to look at the face for IO set in an attempt to name it. Then, the five alternative names were successively presented with an interstimulus interval (1%) of I8 t 3 seconds. Variable ISI’s were used to offset anticipatory autonomic arousal. After all of the names had been successively presented, the patient was given all five at once and was asked to choose the target name from the multiple-choice array. The serial position of the target name was randomly counterbalanced across trials except that it never occurred the first position due to the tendency to respond maximally to the first item in a series (Graham, 1973). Skin conductance data was sampled at 20 Hz for 7 set before (tonic) and 7 set after (stress) presentation of each name. Any change 3 0.02 micromho within 3-5 set of stimulus onset (measured relative to the average tonic SC value) was regarded as an electrodermal response (EDR). Two female subjects matched EDR 0 245 DISCRIMINATION NAMING EDR DISCRIMINATION RECOCNlTlON DEPENDENTMEASURE FIG. 2. Famous face identification (16 trials x 2 subjects). Bars topped (p < .05) using the binomial test. test. Values for controls are based on 32 responses with the same symbol are not significantly different to the patient for age and education neurologic or psychiatric illness. participated as controls. Neither had a history of Results. On each face, three dependent measures were collected. First, the patient’s ability to spontaneously name the face was recorded. Second, the number of times the largest EDR within each series occurred to the target name was tallied. Responses to the first name on each trial, which was never correct, were discarded as suggested by Graham (1973). Third, we tallied the number of times the patient chose the target name during multiple choice. Results are presented in Fig. 2. The patient could not name any of the 16 faces (controls = 98%, p < .OOl), and was only 3 1.25% accurate on selection from multiple choice (controls 98%, p < ,001). In contrast, maximum EDR to the correct name occurred on 62.5% (10/16) of the trials (controls = 40%, n.s.). On 60% of these trials, EDR discrimination of the target name did not lead to correct multiple-choice recognition. We believe that the relatively larger EDR to target names occurs because such names activate stored identity-specific information in memory. One way of further exploring this idea would be to evaluate differential electrodermal responses in a paradigm involving unfamiliar faces with which the patient had no prior contact. If the activation of identityspecific information is important in the EDR discrimination phenomenon, there should be no differential EDR in a matching-to-sample paradigm using unfamiliar faces. We modified the Test of Facial Recognition (Benton et al., 1983) to test this hypothesis. Unfamiliar face recognition rest (UFRT). Ten of the target stimuli from the Benton Test of Facial Recognition were individually mounted on 7.6 x 22.7 cm cards. For each target, five comparison faces were selected, only one of which depicted the target person (this differs from the standard TFR in which, after the fifth item, three of six comparison faces depict the target person on each trial). The task was designed to require the patient to generate a view-independent description of the face to achieve successful completion (cf. Marr, 1982). Thus, on each trial, the comparison face depicted the target person under 246 BAUER AND VERFAELLIE 70 Ii 80 ; 50 fi Iz 40 pL 20 * 30 10 0 DEPENDENT MEASURE FIG. 3. Unfamiliar face recognition test. Values for controls are based on 19 responses (10 trials x 2 subjects, less I trial discarded because of movement artifact). Bars topped with the same symbol are not significantly different @ < .05) using the binomial test. nonisomorphic viewing conditions (i.e., the comparison face was shown with different lighting or angle of view than the target). On each trial. the target face was centrally presented and remained in view throughout the successive presentation of the five comparison faces, which were presented with an interstimulus interval of I8 f 3 sec. The serial position of the correct comparison face was randomly counterbalanced across trials and never occurred in the first position. Skin conductance was sampled as described above. Once the five faces had all been shown, they were arranged in front of the patient who then pointed to the one which depicted the same person as the target. The same two controls who performed the famous faces test also performed the UFRT. Results. Results from the UFRT are presented in Fig. 3. The patient pointed to the correct comparison face on only 30% (3/10) of the trials (controls = 73.68%, p < .05). EDR discrimination was no more accurate than pointing, since the patient showed maximum EDR to the correct comparison face only 30% of the time (controls = 53.63% p < .05). The 30% EDR accuracy shown by the prosopagnosic is not significantly different from the chance probability (25%) of showing a maximum EDR to the target on any given trial. Absolute response magnitudes. Previous data are based on differential response data within each trial and contain no information about the size of the prosopagnosic’s EDRs relative to normals. The average EDR magnitudes to targets and distracters in the FFIT and UFRT are presented in Fig. 4. Inspection of this figure reveals that the prosopagnosic shows extremely small EDR magnitudes compared to normals, even in the FFIT where she clearly differentiated targets from distracters. In all cases, the differences in EDR magnitude between the prosopagnosic patient and the controls are significant at p < .05. DISCUSSION The results from the FFIT confirm previous findings that prosopagnosics can electrodermally discriminate famous faces they cannot identify (Bauer, EDR 247 DISCRIMINATION FAMOUS FACE IDENTIFICATION TEST UNFAMILIAR RECOGNITION FACE TEST FIG. 4. Absolute electrodermal response magnitudes. Dark bars are responses to target names; diagonally striped bars are responses to distractor (foil) names. Bars topped with the same symbol are not significantly different (p < .05). 1984; Tranel & Damasio, 1985). However, the patient was unable to electrodermally discriminate correct from incorrect target-comparison matches in a task using unfamiliar faces. The UFRT, like the FFIT, requires the patient to operate with a view-independent representation of an apprehended face in order to discriminate it from others, but does not require the patient to access the remote memory store to derive identity-specific semantic information important to face identification. Our findings suggest that electrodermal recognition of faces in prosopagnosics exists only when the patient is able to deal with faces that were familiar prior to illness onset. In other words, electrodermal recognition of facial identity in prosopagnosia depends upon activation of remote, not recent, memories in these patients. The patient displayed discriminative electrodermal responses to famous faces she could not visually identify. We believe that such discriminative responses result from the specific presence in memory of a facial representation which “matches” in some important way the percept resulting from visual inspection of the face. The knowledge contained in this representation is “activated” when the correct name is given. However, such activation does not in itself result in conscious verbal identification, either because the “match” is not sufficiently complete, or because the degree of activation so elicited is greatly reduced (Fig. 4) (cf. also Bauer, 1986). Because EDR is most sensitive to the specific pairing of a famous face and its correct name, we believe that in this paradigm such responses reflect the activation of at least a portion of the identity-specific semantic codes described by Bruce and Young (1986). While the nature of the information contained in such codes is still poorly understood, it seems clear that it is neither modality-specific nor dependent upon the idiosyncratic structural characteristics of the face stimulus which activates it. That is, such codes are “viewpoint-independent” (Humphreys & Riddoch, in press). As an example, correct identification of “John F. Kennedy” can normally proceed using any adequate picture of the former President and does not require the presentation of a particular prototype. Thus, the 248 BAUER AND VERFAELLIE information contained in such representations must be of such a nature that it can be activated by an almost unlimited set of specific stimuli. These constructs provide a framework for understanding electrodermal recognition of previously familiar faces. However, when dealing with unfamiliar faces, different cognitive processes may be involved (Bruce & Young, 1986). In the latter case, appeal cannot be made to identityspecific information in memory because the face is being encountered for the first time. Instead, such tasks appear more directly involve the complex perceptual processes by which configurational and semantic information is extracted from the facial array. We believe that that patient’s failure to display differential EDR to correct target-comparison matches in the UFRT results from the fact that she has no stored memory representation of the faces. Before this hypothesis can be accepted, however, alternative accounts of the distinction between the FFIT and UFRT need to be considered. First, it is possible that differential electrodermal responses reflect the operation of early perceptual processes rather than the later process of memory activation. Because the patient does well on isomorphic face matching and shows accurate electrodermal discrimination of correct vs. incorrect face-name pairings on the FFIT, we know that some perceptual abilities remain. However, her poor performance on picture interpretation, reading, and complex matching-to-sample tasks suggests a complex perceptual defect. She is unable to match two faces presented from different viewpoints. Therefore, defects exist relatively “early” in the visual processing chain. It is clear, however, that prosopagnosia can exist in the absence of such profound perceptual defects. In patients with the “associative” form of prosopagnosia, the defect occurs later, perhaps at the stage of processing where stored identity-specific information is accessed by vision. If EDR discrimination in this paradigm reflects perceptual factors, then patients with associative forms of prosopagnosia would show better electrodermal discrimination of correct and incorrect matches on the matching-to-sample paradigm. Studies are currently underway in our laboratory to examine this possibility. Second, stimulus or task factors could be partially responsible for our findings. It may be, for example, that the face stimuli in the UFRT were more complex or somehow different in their processing requirements than those in the FFIT. Face stimuli used in the UFRT were photographed under poor lighting conditions or at oblique angles, and thus possibly differ from FFIT faces in number of available features, spatial frequency characteristics, or other variables shown to be important for face processing in previous research (cf. Sergent, 1986). There is evidence to suggest that matching tasks involve the use of different facial attributes than do identification tasks (Ellis et al., 1979), and may even require the use of different spatial frequencies (Ginsburg, 1978; Fiorentini, Maffei, & Sandini, EDR 249 DISCRIMINATION 1983; Sargent, 1986). Finally, it is noteworthy that the FFIT was an auditory-visual task (names matched with faces), while the UFRT involved a visual-visual match. Currently, we are modifying these tasks for use with prosopagnosics in order to more fully understand the role of stimulus and task factors in the autonomic recognition phenomenon. One way to make the two tasks more comparable would be to construct a famous face recognition test in which the patient was required to match famous persons depicted at different photographic angles.’ If our “memory” interpretation of autonomic recognition is correct, then agnosic patients who are able to extract “viewpoint-independent” descriptions of faces should show differential electrodermal responses to correct matches in this paradigm, but not on the UFRT. Our contention that electrodermal recognition of faces does not occur for faces first encountered since illness onset contradicts the findings of a recent study by Tranel and Damasio (1985). Their second patient showed larger EDR to faces learned since illness onset than to unfamiliar faces. This patient became unable to learn new faces after a herpes simplex encephalitis resulted in bilateral posterior temporal lobe lesions, but importantly had no difJiculty recognizing previously.familiar persons. Tranel and Damasio refer to the patient’s deficit as an “anterograde prosopagnosia” (p. 228). We would argue that the discrepancy between our results and Tranel and Damasio’s Patient 2 occurs because the latter patient is not classically prosopagnosic. Instead, her deficit may be more properly understood as a visual recent memory disorder, which Ross (1980) has shown can be dissociated from true prosopagnosia. Tranel and Damasio’s data suggest that amnesics can demonstrate new learning in the autonomic channel which is not reflected in verbal report, which is consistent with a large body of data on memory without awareness in amnesic patients (Cohen, 1984). In contrast, we are unaware of any patient with welldefined prosopagnosia [total (retrograde and anterograde) loss in the ability to identify faces] who can learn new faces, regardless of whether such learning is measured via verbal or electrodermal responses. We suggested earlier that electrodermal orienting to the correct name in the FFIT reflects the activation of identity-specific information linking face and name. Although some aspect of memory is involved, it is still unclear what part of the memory process is indexed by the electrodermal response. For example, is the EDR sensitive to memory search or does it reflect the actual retrieval of identity-specific information from the memory store? Our original hypothesis, based on the rationale underlying the use of the Guilty Knowledge Test in police interrogation (Lykken, 1959), was that the EDR represented recognition of the most significant piece of information (the correct face-name match) in the multiple-choice I We thank an anonymous reviewer for suggesting this approach. 250 BAUER AND VERFAELLIE array. This is essentially a retrieval explanation. Although this hypothesis may have some merit, it does not explain those instances where normals are able to identify faces without displaying differential EDR to the correct face-name match (see Figs. 2 and 3). At the time our original hypothesis was advanced, little guidance was available which might link concepts of “orienting” to “memory,” “attention,” “ significance detection,” etc. Recently, however, an information processing model of orienting has been offered by Ohman (1979) which may be helpful in understanding the autonomic recognition phenomenon. In Ohman’s model, large EDR’s reflect not only access to the memory system, but also the effortful processing of sensory information when it does not immediately match stored information in memory. Such processing allegedly takes place in a central limited-capacity channel which involves focal attention. We believe that normal controls can identify faces without large EDR because famous face identification is largely an automatic process (Hay, Young, & Ellis, 1986). This viewpoint agrees with findings from a preliminary analysis which indicated that discriminative accuracy of the EDR improves under conditions of reduced confidence. That is, normals appeared to show greater and more consistent EDR discrimination of the correct name (FFIT) or comparison face (UFRT) when they were unsure of their choices, and showed practically no differential EDR when they were immediately certain of the identity of a face they were viewing. This suggests that differential EDR depends on the degree of effort required to perform the identification task and on whether the matching representation is readily available in memory. In prosopagnosics, the automatic components of face identification no longer operate. We propose that prosopagnosia involves a failure of an incoming percept to match a memory representation with sufficient completeness or certainty to result in positive overt identification. However, at some more basic level, the stimulus is defined as having greater significance than the others which surround it. Thus, more attentional effort is spent toward processing it, and this results in a larger EDR to the correct face-name pair. The hypothesis that autonomic recognition is dependent upon previously stored identity-specific information has implications for the study of new learning of facial information by prosopagnosics. On the one hand, it could be predicted that prosopagnosics would be unable to learn new facial identity information because the causative lesions destory the neural substrate for object and face identification in the ventral visual-limbic pathway (Bauer, 1984). On the other hand, such lesions typically spare the dorsal visual-limbic pathway which is important in detecting some aspects of stimulus significance. Because this results in some sparing of visual-limbic connections, repeated exposure to faces might lead to the EDR DISCRIMINATION 251 development of autonomic discrimination even though such effects are not manifested in verbal report. We were unable to perform this test on the current patient because she moved to a distant state on discharge from the hospital. However, an experiment which examines verbal and electrodermal aspects of the familiarization process in a second prosopagnosic is currently underway in our laboratory. REFERENCES Albert, M. S., Butters, N., & Levin. J. 1979. Temporal gradients in the retrograde amnesia of patients with alcoholic Korsakoff’s disease. Archives of Neurology, 36, 21 I-216. Bauer, R. M. 1984. Autonomic recognition of names and faces in prosopagnosia: A neuropsychological application of the Guilty Knowledge Test. Neltropsy~koloRitr, 22,457469. Bauer, R. M. 1986. The cognitive psychophysiology of prosopagnosia. In H. D. Ellis, M. A. Jeeves, F. Newcombe. & A. W. Young (Eds.). Aspects of’firce proce.ssing (NATO ISI Series No. 38). Dordrecht, UK: Nijhoff. Bauer, R. M., & Rubens, A. B. 1985. Agnosia. In K. M. Heilman & E. Valenstein (Eds.). Clinicul neuvopsychalo~y. New York: Oxford Univ. Press. 2nd ed. Bauer, R. M., & Trobe, J. R. 1984. Visual memory and perceptual impairments in prosopagnosia. Joarnal of Clinical Nerrro-opl~thal~nolo~~. 4, 39-46. Benton, A. L. 1980. The neuropsychology of face recognition. American Psychologist, 3.5, 176-186. Benton, A. L., Hamsher, K., Varney, N. R.. & Spreen. 0. 1983. Contrihation.s to net,ropsychologicul assessment. New York: Oxford Univ. Press. Bodamer, J. 1947. Prosopagnosie. Arch Psychiatr Nerve&r, 179, 6-54. Bruce, V., & Young, A. 1986. Understanding face recognition. British Jorrrnal ofP.~ycholo~~, 77, 305-327. Cohen, N. J. 1984. Preserved learning capacity in amnesia: Evidence for multiple memory systems. In L. R. Squire & N. Butters (Eds.), Nf~r,rop.s~c,ho/oRy of‘ memory. New York: Guilford Press. Damasio, A. R.. Damasio, H., & Van Hoesen, G. W. 1982. Prosopagnosia: Anatomic basis and behavioral mechanisms. Neurology, 32, 331-341. Delis, D. C., Kramer, J., Kaplan, E. F., & Ober. B. A. 1983. California \,erha/ letrrnitlg test. Martinez, CA: Life Science Associates. Ellis, H. D., Shepherd, J. W., & Davies, G. M. 1979. Identification offamiliar and unfamiliar faces from internal and external features: Some implications for theories of face recognition. Perception, 8, 431-439. Fiorentini, A., Maffei, L., & Sandini. G. 1983. The role of high spatial frequencies in face perception. Perception, 12, 195-201. Ginsburg, A. P. 1978. Visual information processing based on spatial filters constrained by hiologiral duta (AMRL Technical Report 78-129). Ohio. Graham, F. K. 1973. Habituation and dishabituation of responses innervated by the autonomic nervous system. In H. V. S. Peeke & M. J. Herz (Eds.), Hubitaation: Behavioral studies. New York: Academic Press. Vol. 1. Hay, D. C., Young, A. W., & Ellis, A. W. 1986. What happens when a face rings a bell?: The automatic processing of famous faces. In H. D. Ellis, M. A. Jeeves, F. Newcombe & A. W. Young (Eds.). Aspects of,fuc,e processing (NATO IS1 Series No. 38). Dordrecht, UK: Nijhoff. Humphreys, G. W., & Riddoch, M. J. (in press). The fractionation of visual agnosia. In G. W. Humphreys & M. J. Riddoch (Eds.), Visd object processing: A cognitilze nel~rp.~yc~~lnlo~ic~~l rrpprorrr,h. London: Erlbaum. 252 BAUER AND VERFAELLIE Luria, A. R. 1959. Disorders of “simultaneous perception” in a case of bilateral occipitoparietal brain injury. Brain, 83, 437-449. Lykken, D. T. 1959. The GSR in the detection of guilt. Journal of Applied Psychology. 43, 385-388. Malone, D. R. Morris, H. H., Kay, M. C., & Levitt, H. S. 1982. Prosopagnosia: A double dissociation between the recognition of familiar and unfamiliar faces. Journal of Neurology, Neurosurgery, and Psychiatry, 45, 820-822. Marr, D. 1982. Vision. San Francisco: Freeman. Ohman, A. 1979. The orienting response, attention, and learning: An information-processing perspective. In H. Kimmel, E. Van Olst & J. Orlebeke (Eds.), The orienting reflex in humans. Hillsdale, NJ: Erlbaum. Ross, E. D. 1980. Sensory-specific and fractional disorders of recent memory in man. 1. Isolated loss of visual recent memory. Archives of Neurology, 37, 193-200. Sergent, J. 1986. Microgenesis of face perception. In H. D. Ellis, M. A. Jeeves, F. Newcombe, & A. W. Young (Eds.), Aspects offace processing (NATO IS1 Series No. 38). Dordrecht, UK: Nijhoff. Tranel, D., & Damasio, A. R. 1985. Knowledge without awareness: An autonomic index of facial recognition by prosopagnosics. Science, 228, 1453-1454. Warrington, E. K., & James, M. 1967. An experimental investigation of facial recognition in patients with unilateral cerebral lesions. Cortex, 3, 317-326. Wolpert, I. 1924. Die simultanagnosie: Storung der gesamtauffassung. Zeifschrift gesamte Neural Psychiat, 93, 397-413. Young, A. W., Hay, D. C., & Ellis, A. W. 1986. Getting semantic information from familiar faces. In H. D. Ellis, M. A. Jeeves, F. Newcombe, & A. W. Young (Eds.), Aspects of face processing. Dordrecht, UK: Nijhoff.