Is the face-sensitive N170 the only ERP not affected by selective attention?

Transcription

1 COGNITIVE NEUROSCIENCE Is the face-sensitive N17 the only ERP not affected by selective attention? Alexandra SeÂverac Cauquil, CA Gillian E. Edmonds 1 and Margot J. Taylor Centre de Recherche Cerveau et Cognition, UMR CNRS/UPS 5549 Universite Paul Sabatier, 133 route de Narbonne, 3162 Toulouse cedex 4, France; 1 Hospital for Sick Children, Faculty of Medicine, University of Toronto, Toronto, Canada CA Corresponding Author Received 3 April 2; accepted 27 April 2 We assessed the effect of directed attention on early neurophysiological indices of face processing, measuring the N17 event-related potential (ERP). Twelve subjects were tested on two tasks each in which they attended either to eyes only or to faces with eyes closed, presented within series of facial and control stimuli. Consistent with the ERP literature, N17 was recorded to facial stimuli at posterior temporal electrodes and a concomitant positive peak at the vertex, with latencies around 15 ms for faces and 174 ms for eyes. However, unlike fmri studies, neither the latency nor the amplitude of the peaks were sensitive to the target/non-target status of either the eyes or the face stimuli. This suggests that early stages of face processing indexed by N17 are automatic and unmodi ed by selective attention. NeuroReport 11:2167± 2171 & 2 Lippincott Williams & Wilkins. Key words: Event-related potential (ERP); Face processing; Selective attention INTRODUCTION Numerous studies have shown that directed attention in uences processing in visual tasks, on both behavioural tests and a range of neuroimaging techniques. Eventrelated potential (ERP) research has demonstrated that selective attention is re ected in very early stages of information processing [1,2]. For spatial attention tasks, which have been the most studied, attention is directed covertly to a spatial location and the resultant peaks between 1 and 3 ms are enhanced, without any change in latencies [3]. For visual features such as colour or shape, attention is usually manipulated by de ning some feature(s) or category as a target, to which the observers must respond, and these attention processes also show clear, early effects on the recorded ERPs either in amplitude or latency to the target compared to the non-target stimuli [4± 7]. Presentation of faces evokes a negative peak at 15± 18 ms (labelled N17) at posterior temporal sites and re ects early visual processing responsible for the encoding of face-speci c visual invariants and forming a sensory representation of the human face [8]. N17 has longer latencies to unusual facial representations, such as distorted or inverted faces [9±11] and to eyes presented alone. Images of faces also evoke a positive peak at the level of the midline parieto-central region of the scalp [12], the vertex positive peak (VPP), although it does not appear to have the same face sensitivity as N17 [13,14]. The effect of directed attention on early face processing has been investigated with fmri, and some studies reported increased activation with covert attention to faces in the fusiform gyrus [15±17], whereas others failed to show consistent changes during attention to faces [18]. In a PET study, Haxby et al. [19] reported increased blood ow in the fusiform gyrus with a face matching task. Neurophysiological studies of selective attention to faces have used only recognition or higher-order categorization tasks [2,21]. In the present study we address the question of whether attentional processes in uence the face-sensitive N17, as the in uence of attention on early stages of face encoding have not been investigated. The effect of attention was studied using a simple classi cation task, with two different categories of targets chosen from the facial stimuli used: faces with eyes closed or eyes only stimuli. MATERIALS AND METHODS Subjects: Twelve subjects were tested (mean age years, four females), recruited from hospital staff and university students. All subjects were right-handed, had normal or corrected to normal vision. Visual stimuli: Grey scale photographs were presented in the following categories with 75 exemplars in each: upright faces with eyes forward, with eyes closed, and with eyes averted, inverted faces, phase-scrambled faces, eyes only, lips only and owers. The target stimuli were either eyes only or faces with eyes closed; subjects were instructed to make a button press to targets. Therefore attention was directed, according to the target, to either faces with eyes closed or to eyes. Stimuli were presented for 4 ms with & Lippincott Williams & Wilkins Vol 11 No 1 14 July

2 A. SE Â VERAC CAUQUIL, G. E. EDMONDS AND M. J. TAYLOR an ISI of 1.8±2.2 s. Stimuli were randomly ordered and in four blocks of 15 stimuli each. Two blocks were selected in which eyes were targets and two in which faces with eyes closed were targets. Short pauses of a few minutes were given between trial blocks; the order in which the pairs of trial blocks with the eyes or the faces with eyes closed served as targets was varied across subjects. ERP recording: ERPs were recorded from 29 electrodes applied with an ECI electrode cap in the 1-1 system. ERPs were recorded with a NeuroScan system 3.1 for 1 s, with a 5 ms prestimulus baseline, with a bandpass of.1± 1 Hz. An averaged reference montage was used, with Cz as the reference lead, and the averaged reference calculated off-line. Trials were rejected for EOG or movement artefact (. 15 ìv), baseline corrected and then averaged according to the stimulus categories. The components of interest were N17 measured at the posterior-temporal electrodes, P7, P8,, and P1, and VPP at the electrode. The peaks were measured within a window of 125±225 ms. For data analyses, analyses of variance with repeated measures were conducted for three categories of facial stimuli (faces with eyes forward, faces with eyes closed, and eyes) and the ower stimulus across the two measures (amplitude and latencies). The latency and amplitude data for the N17 and VPP were analysed by stimuli (four levels), target condition (two levels) and, for the N17, hemisphere (two levels). Additional paired t-tests were performed to test the effect of attention. RESULTS As classically shown in the literature, a clear negative component, N17, was observed on the individual average recordings from the posterior temporal electrodes (P7, P8,, P1) for the facial stimuli (Fig. 1). As seen in other series, the eye stimuli evoked the largest N17, independent of target condition, being signi cantly larger than faces with eyes closed and longer in latency. The ower stimulus produced a much smaller N17, often above baseline. The VPP, measured at the anterior electrode, showed similar effects. Posterior temporal electrodes: Only the nature of the stimulus in uenced N17 latency (F 1,9 ˆ ; p,.1); similar latencies were found regardless of the hemisphere or the target condition (F 1,11 ˆ.264; p ˆ.618 and F 1,11 ˆ.382; p ˆ.549, respectively, no interaction). Face stimuli yielded the shorter latencies (faces with eyes forward: 148 ms and faces with eyes closed: 151 ms) and eyes the longest (174 ms), the latency of the ERP produced in response to owers was slightly below but not signi cantly different than eyes (17 ms). Stimulus was found to be the most in uential factor on the amplitude of the N17 response (F 1,9 ˆ , p,.1). N17 was larger over the right hemisphere sites and larger at inferior (, P1) than superior (P7, P8) sites (F 1,11 ˆ 5.526; p ˆ.38; P7: 4.1 ìv; : 6.7 ìv; P8: 6.1 ìv; P1: 7.7 ìv). The target condition was not signi cant on the global ANOVA (F 1,11 ˆ 3.535; p ˆ.87), although a slight trend was observed on the recordings. As illustrated in Fig. 2, selective attention tended to enhance the response for the eye stimuli whereas it had the opposite effect on face stimuli. That opposite trend could explain the lack of signi cance on the global ANOVA, but an additional paired t-test analysis revealed that only the response evoked by the eye stimuli was larger when eyes were the target, and this at P7 only (t 11 ˆ 2.516; p ˆ.29). Difference potentials (stimuli when target minus the same stimuli when non-target) were also examined, and no signi cant effects were found. VPP: As found for N17, the VPP latency was signi cantly longer for the eye stimuli than for the face stimuli (176 ms vs 15 ms, F 3,9 ˆ 3.194; p,.1). VPP latency for the ower stimuli was situated in between (162 ms). The target condition did not in uence this parameter (F 1,11 ˆ 1.554; p ˆ.238), i.e. latency did not change for the eye stimuli or the faces with eyes closed stimuli as a function of whether they were, or were not, targets. The amplitude of the VPP component depended on the stimulus (F 3,9 ˆ 1.52; p,.1) but not on the target condition (F 1,11 ˆ.646; p ˆ.439) in spite of the tendency, as observed for posterior temporal electrodes, for eyes stimuli to yield a larger response when attended (Fig. 2). The response to face stimuli was very similar regardless of whether they were or were not targets. The eye stimuli produced the largest response (4.9 ìv) and owers the smallest (1 ìv). Face stimuli produced responses with very similar and intermediate amplitudes (2.9 ìv, faces eyes forward, and 3.1 ìv faces, eyes closed). To ensure our lack of signi cance was not due to general attention effect as subjects paid attention to faces, and thus all facial stimuli were more attended and enhanced, we compared the present data to adult data (n ˆ 38) from a previous study which had included the same stimulus categories, but in which the target stimulus was a checkerboard. We found no signi cant differences between N17 to the faces with eyes closed measured in the current study regardless of target/non target status and the same stimuli when measured in a series when checkerboards were targets (F 1,192 ˆ 2.35, p ˆ.127). Likewise, there were no differences between the eyes when non-targets in the present study compared to eyes in the earlier study when checkerboards were targets (F 1,192 ˆ1.338, p ˆ.249). DISCUSSION Whereas most of the visual processes appear to be sensitive to directed attention, the present study found that early face processing presents the striking particularity to remain unchanged by selective attention. In fact there was a surprising tendency: when eyes were the target, eye stimuli tended to produce a larger N17, but face stimuli lead to a smaller N17 when they were attended. The one attentional effect, for the eyes stimuli only, was seen over the left hemisphere, and at the more superior electrode site, where the response is minimal. It could be hypothesized that at the other locations, response is not in uenced by attention as it is already at its maximal level. Other ERP studies that have directed attention to face stimuli have required higher level processing than the present study, such as decisions regarding familiarity, gender, recognition memory or famous/non-famous categorisation, and none reported any effects on N17 [2±22] but consistently effects that re ect later stages of proces Vol 11 No 1 14 July 2

3 FACE-SENSITIVE N17 AND SELECTIVE ATTENTION faces, eyes closed, targets eyes targets P1 P eyes faces, eyes open faces, eyes closed flowers Fig. 1. Grand averages (n ˆ 12) of the ERPs to the four categories of stimuli: eyes only, faces with eyes open, faces with eyes closed, owers, at the, and P1 electrodes. sing. All of the fmri studies that have shown increased activation to faces have used paradigms that required a memory component (1-back or recognition tasks [15,17,18]), and it has been well documented in the ERP literature that the stages of processing required for such tasks occur at longer latencies than N17. For example, Clark et al. [18] found greater activation to a single target face than to novel faces in the lateral temporal cortex, whereas the opposite pattern was seen in the medial area of the ventral temporal cortex. They explained this effect as due to the segregation of the activation as a function of the novelty of the stimuli. They used a memory task, as one target face was learned and presented rarely as a target. In the present study, no facial stimuli were repeated either as targets or non-targets, and thus all were novel, and the stimulus categories were also Vol 11 No 1 14 July

4 A. SE Â VERAC CAUQUIL, G. E. EDMONDS AND M. J. TAYLOR eyes, attended eyes, non-attended faces, non-attended faces, attended P7 P8 P1 Fig. 2. Mean and s.e.m. of the amplitude of VPP (measured at ) and N17 (measured at P7, P8, and P1) evoked by eyes and faces with eyes closed stimuli, when attended and not attended. equiprobable. These differences in methodology could clearly explain the differences between the nding of the two studies, apart from the issue discussed above of the longer latency time resolution in fmri work. With the high time resolution of ERPS, we can measure very early stages of cognitive processing and these were not affected by selective attention to faces. This was nevertheless unexpected as many studies have shown attentional effects in latency and amplitude on very early stages of processing, in both visual [5] and auditory [1] modalities. However, intracranial studies of the N2 face-speci c component have found that it too is not sensitive to a wide range of manipulations expected to produce top-down in uences [23]. It may be that due to the importance of faces, the early processing of facial stimuli is requisite and automatic, and only subsequent stages are amenable to attentional in uences. CONCLUSION This work fails to show any enhancement of the early neurophysiological response to faces due to attention, in contrast with recent fmri studies that have reported that the signal evoked by faces was increased by attention. This difference is likely due to the fact that fmri or PET activations are measured over a much longer period than the ERPs in the current work. Another factor is the difference in paradigms. In the fmri and PET studies, the subjects were involved in recognition tasks, and numerous neurophysiological studies have demonstrated that recognition of faces produces effects at longer latencies than the N17 [2]. Thus, the N17 to face stimuli does not change, regardless of the target stimuli, be they faces, other facial stimuli such as eyes, or non-facial stimuli as checkerboards; our lack of effect in the present study was not due to a general arousal/attentional effect when only face stimuli were targets. This suggests that N17 indexes to a larger extent an automatic processing of facial stimuli, resistant to attentional effects, which in uence other activity in various brain regions, but at longer latencies, as clearly shown in intracranial recordings [23,24]. REFERENCES 1. Hillyard SA and Anllo-Vento L. Proc Natl Acad Sci USA 95, 781± Wijers AA, Mulder G, Gunter TC et al. Brain potentials analysis of selective attention. In: Handbook of Perception and Action. New York: Academic Press, 1996: 333± Mangun GR and Hillyard SA. Electrophysiological studies of visual selective attention in humans. In: Scheibel AB and Wechsler A (eds). The Neurobiological Foundations of Higher Cognitive Function. New York: Guilford Press, 1998: 271± Han S, Fan S, Chen L et al. J Cogn Neurosci 9, 687±698 (1997). 5. Anllo-Vento L, Luck SJ and Hillyard SA. Human Brain Mapp 6, 216± Lange JJ, Wijers AA, Mulder LJM et al. Biol Psychology 48, 153± Vol 11 No 1 14 July 2

5 FACE-SENSITIVE N17 AND SELECTIVE ATTENTION 7. Taylor MJ and Khan SC. Int J Psychophysiol 2, in press. 8. Bentin S, Allison T, Puce A et al. J Cogn Neurosci 8, 551±565 (1996). 9. George N, Evans J, Fiori N et al. Cogn Brain Res 4, 65±76 (1996). 1. Eimer M. NeuroReport 9, 2945± Rossion B, Delvenne JF, Debatisse D et al. Biol Psychol 5, 173±189 (1999). 12. Jeffreys DA. Exp Brain Res 96, 163±172 (1993). 13. Bbtzel K, Schulze S and Stodieck SRG. Exp Brain Res 14, 135±143 (1995). 14. Schendan HE, Ganis G and Kutas M. Psychophysiology 35, 24± Wojciulik E, Kanwisher N and Driver J. J Neurophysiol 79, 1574± O'Craven KM, Downing PE and Kanwisher N. Nature 41, 584±587 (1999). 17. Shouten JL, Furey ML and Haxby JV. Soc Neurosci Abstr 25, 163 (1999). 18. Clark VP, Parasuraman R, Keil Y et al. Hum Brain Mapp 5, 293±297 (1997). 19. Haxby JV, Horwitz B, Ungerleider LG et al. J Neurosci 14, 6336±6353 (1994). 2. Sommer W, Schweinberger SR and Matt J. Electroencephalogr Clin Neurophysio 79, 457±463 (1991). 21. Mouchetant-Rostaing Y, Giard MH, Bentin S et al. Eur J Neurosci 12, 33±31 (2). 22. Barrett SE, Rugg MD and Perrett DI. Neuropsychologia 26, 15±117 (1988) 23. Puce A, Allison T and McCarthy G. Cerebr Cortex 9, 445±458 (1999). 24. McCarthy G, Puce A, Belger A et al. Cerebr Cortex 9, 431±444 (1999). Vol 11 No 1 14 July