Visual perceptual learning in human object recognition areas: a repetition priming study using high-density electrical mapping.
ABSTRACT It is often the case that only partial or degraded views of an object are available to an observer, and yet in many of these cases, object recognition is accomplished with surprising ease. The perceptual filling-in or "closure" that makes this possible has been linked to a group of object recognition areas in the human brain, the lateral occipital (LO) complex, and has been shown to have a specific electrophysiological correlate, the N(cl) component of the event related potential. Perceptual closure presumably occurs because repeated and varied exposure to different classes of objects has caused the brain to undergo "perceptual learning," which promotes a robust mnemonic representation, accessible under partial information circumstances. The present study examined the impact of perceptual learning on closure-related brain processes. Fragmented pictures of common objects were presented, such that information content was incrementally increased until just enough information was present to permit closure and object recognition. Periodic repetition of a subset of these picture sequences was used to induce repetition priming due to perceptual learning. This priming has an electrophysiological signature that is putatively generated in the LO complex, but significantly precedes the electrophysiological correlate of closure. The temporal progression of priming- and closure-related activity in the LO complex supports the view that sensory processing entails multiple reentrant stages of activity within processing modules of the visual hierarchy. That the earliest priming-related activity occurs over LO complex, suggests that the sensory trace itself may reside in these object recognition areas.
- [Show abstract] [Hide abstract]
ABSTRACT: The way we perceive an object depends both on feedforward, bottom-up processing of its physical stimulus properties and on top-down factors such as attention, context, expectation, and task relevance. Here we compared neural activity elicited by varying perceptions of the same physical image-a bistable moving image in which perception spontaneously alternates between dissociated fragments and a single, unified object. A time-frequency analysis of EEG changes associated with the perceptual switch from object to fragment and vice versa revealed a greater decrease in alpha (8-12 Hz) accompanying the switch to object percept than to fragment percept. Recordings of event-related potentials elicited by irrelevant probes superimposed on the moving image revealed an enhanced positivity between 184 and 212 ms when the probes were contained within the boundaries of the perceived unitary object. The topography of the positivity (P2) in this latency range elicited by probes during object perception was distinct from the topography elicited by probes during fragment perception, suggesting that the neural processing of probes differed as a function of perceptual state. Two source localization algorithms estimated the neural generator of this object-related difference to lie in the lateral occipital cortex, a region long associated with object perception. These data suggest that perceived objects attract attention, incorporate visual elements occurring within their boundaries into unified object representations, and enhance the visual processing of elements occurring within their boundaries. Importantly, the perceived object in this case emerged as a function of the fluctuating perceptual state of the viewer.Journal of Vision 07/2013; 13(13). · 2.48 Impact Factor
- [Show abstract] [Hide abstract]
ABSTRACT: WE LIVE IN A CLUTTERED, DYNAMIC VISUAL ENVIRONMENT THAT POSES A CHALLENGE FOR THE VISUAL SYSTEM: for objects, including those that move about, to be perceived, information specifying those objects must be integrated over space and over time. Does a single, omnibus mechanism perform this grouping operation, or does grouping depend on separate processes specialized for different feature aspects of the object? To address this question, we tested a large group of healthy young adults on their abilities to perceive static fragmented figures embedded in noise and to perceive dynamic point-light biological motion figures embedded in dynamic noise. There were indeed substantial individual differences in performance on both tasks, but none of the statistical tests we applied to this data set uncovered a significant correlation between those performance measures. These results suggest that the two tasks, despite their superficial similarity, require different segmentation and grouping processes that are largely unrelated to one another. Whether those processes are embodied in distinct neural mechanisms remains an open question.Frontiers in Psychology 01/2013; 4:795. · 2.80 Impact Factor
- [Show abstract] [Hide abstract]
ABSTRACT: The adult human visual system can efficiently fill-in missing object boundaries when low-level information from the retina is incomplete, but little is known about how these processes develop across childhood. A decade of visual-evoked potential (VEP) studies has produced a theoretical model identifying distinct phases of contour completion in adults. The first, termed a perceptual phase, occurs from approximately 100-200ms and is associated with automatic boundary completion. The second is termed a conceptual phase occurring between 230-400ms. The latter has been associated with the analysis of ambiguous objects which seem to require more effort to complete. The electrophysiological markers of these phases have both been localized to the lateral occipital complex, a cluster of ventral visual stream brain regions associated with object-processing. We presented Kanizsa-type illusory contour stimuli, often used for exploring contour completion processes, to neurotypical persons ages 6-31 (N=63), while parametrically varying the spatial extent of these induced contours, in order to better understand how filling-in processes develop across childhood and adolescence. Our results suggest that, while adults complete contour boundaries in a single discrete period during the automatic perceptual phase, children display an immature response pattern - engaging in more protracted processing across both timeframes and appearing to recruit more widely distributed regions which resemble those evoked during adult processing of higher-order ambiguous figures. However, children older than 5years of age were remarkably like adults in that the effects of contour processing were invariant to manipulation of contour extent.NeuroImage 12/2013; · 6.25 Impact Factor
Visual Perceptual Learning in Human Object Recognition Areas:
A Repetition Priming Study Using High-Density Electrical Mapping
Glen M. Doniger,*,† J ohn J . Foxe,*,‡,§,1Charles E. Schroeder,*,§ Micah M. Murray,*,§
Beth A. Higgins,* and Daniel C. J avitt*,‡,§
*Cognitive Neurophysiology Laboratory, Cognitive Neuroscience and Schizophrenia Program, Nathan Kline Institute for Psychiatric
Research; †Department of Psychology, New York University; ‡Department of Psychiatry and Behavioural Sciences and
§Department of Neuroscience, Albert Einstein College of Medicine
Received J une 13, 2000; published online December 21, 2000
It is often the case that only partial or degraded
views of an object are available to an observer, and yet
in many of these cases, object recognition is accom-
plished with surprising ease. T he perceptual filling-in
or “closure” that makes this possible has been linked
to a group of object recognition areas in the human
brain, the lateral occipital (L O) complex, and has been
shown to have a specific electrophysiological corre-
late, the Nclcomponent of the event related potential.
Perceptual closure presumably occurs because re-
peated and varied exposure to different classes of ob-
jects has caused the brain to undergo “perceptual
learning,” which promotes a robust mnemonic repre-
sentation, accessible under partial information cir-
cumstances. T he present study examined the impact
of perceptual learning on closure-related brain pro-
cesses. F ragmented pictures of common objects were
presented, such that information content was incre-
mentally increased until just enough information was
present to permit closure and object recognition. Pe-
riodic repetition of a subset of these picture sequences
was used to induce repetition priming due to percep-
tual learning. T his priming has an electrophysiologi-
cal signature that is putatively generated in the L O
complex, but significantly precedes the electrophysio-
logical correlate of closure. T he temporal progression
of priming- and closure-related activity in the L O com-
plex supports the view that sensory processing entails
multiple reentrant stages of activity within processing
modules of the visual hierarchy. T hat the earliest
priming-related activity occurs over L O complex, sug-
gests that the sensory trace itself may reside in these
object recognition areas.
© 2001 Academic Press
A striking feature of the human visual system is its
ability to recognize objects from degraded or incom-
plete views (e.g., Snodgrass and Feenan, 1990). The
term “perceptual closure” has been used todescribe the
apparent filling-in of missing information that enables
object-recognition under such partial viewing condi-
tions (e.g., Bartlett, 1916; Foley et al., 1997). A recent
electrophysiological study showed that perceptual clo-
sure processes are linked to differential activation of
thelateral occipital (LO) complex (Doniger et al., 2000),
a system of areas implicated in object recognition in
humans (e.g., Malach et al., 1995; Kanwisher et al.,
1997; Grill-Spector et al., 1999; Haxby et al., 1999), and
considered analogous or even homologous with infero-
temporal (IT) cortices in macaque monkeys (e.g., Sary
et al., 1993; Ito et al., 1995; Vogels, 1999). In our pre-
vious study (Doniger et al., 2000), high-density electri-
cal mapping revealed a robust event-related potential
component that appeared totrack the neural processes
related to perceptual closure (termed Nclfor negativity
associated with closure). This component was seen as a
relative negativity over bilateral
scalp (onset ?230 ms, peak latency at ?290 ms) and
was maximal when just enough visual information was
present for subjects to finally identify fragmented im-
Object recognition under incomplete viewing condi-
tions is presumably possible because previous expo-
sure to the object and like objects of its class, under a
variety of viewing conditions (e.g., viewing angles,
sizes, lighting conditions), has resulted in perceptual
learning such that the appropriate mnemonic repre-
sentation can be accessed despite the impoverished
information. In fact, it has been shown that only brief
1To whom correspondence and reprint requests should be ad-
dressed. Cognitive Neurophysiology Laboratory, Nathan Kline Insti-
tute for Psychiatric Research, Program in Cognitive Neuroscience
and Schizophrenia, 140 Old Orangeburg Road, Orangeburg, NY
10962, USA. Fax: (845) 398-6545. E-mail: firstname.lastname@example.org.
NeuroImage 13, 305–313 (2001)
doi:10.1006/nimg.2000.0684, available online at http://www.idealibrary.com on
Copyright © 2001 by Academic Press
All rights of reproduction in any form reserved.
exposure to a complete or unambiguous object is re-
quired for perceptual learning to result in repetition
priming, enabling recognition of a previously ambigu-
ous version of thesameobject (Toveeet al., 1996; Dolan
et al., 1997).
This study uses high-density electrical mapping to
examine the effect of repetition priming due to percep-
tual learning on closure-related activity over the LO
complex. Our paradigm involved presenting sequences
of fragmented pictures such that incrementally more
complete versions of an object are presented until just
enough information is present for subjects to “close”
the picture and recognize the object (e.g., Snodgrass
and Corwin, 1988; Doniger et al., 2000). Priming due to
perceptual learning was induced by the repetition of a
picturesequencefollowing its initial presentation, with
one or two novel picture sequences intervening. As
expected, perceptual learning enabled recognition to
occur at a significantly more fragmented level when
picture sequences were repeated as compared to when
initially presented. For example, the stimuli shown in
Fig. 1c were uninterpretable to most of our subjects
when initially presented, but became interpretable fol-
lowing perception of the complete versions of the stim-
uli (shown in Fig. 7). Moreover, two successive en-
hancements in electrical activity over LO were found
for highly fragmented images that were unrecogniz-
able when initially presented, but became recognizable
when repeated due to perceptual learning. The later
enhancement is consistent with our prior finding of a
component of the evoked potential (Ncl) that is most
pronounced when closure is possible and object recog-
nition is achieved (Doniger et al., 2000). We propose
that the earlier enhancement is an electrical signature
of priming due toperceptual learning, reflecting access
to a sensory trace laid down when the object was ini-
tially recognized. The temporal progression from prim-
ing- to closure-related activity over LO supports the
view that sensory perceptual processing entails multi-
ple reentrant stages of activity within modules of the
visual hierarchy (e.g., Schroeder et al., 1998). That the
earliest priming-related activity occurs over the hu-
man LO complex, suggests that the sensory trace itself
resides in these object-recognition areas.
MATERIALS AND METHODS
Ten (3 female), neurologically normal, paid volun-
teers, aged 20–30 (mean ? 23.6) participated. All sub-
jects provided written informed consent, and the pro-
cedures were approved by the Institutional Review
Board of the Nathan Kline Institute. All subjects re-
ported normal or corrected-to-normal vision. Nine of
the 10 were right-handed.
Stimuli and Task
Subjects were presented with 400 line drawings
(black on a gray background) of natural and man-made
objects; 260 from the Snodgrass and Vanderwart
(1980) normed set; the rest from Cycowicz et al. (1997).
Images were 256 ? 256 pixel bitmaps, divided into
16 ? 16 segments. Segments containing black pixels
were randomly and cumulatively deleted to produce
seven incrementally fragmented versions of each pic-
ture (Snodgrass and Corwin, 1988). Level 1 refers to
the complete picture and Level 7 to the most frag-
mented version, where the proportion of deleted seg-
ments for any level equals [1–0.7(level-1)]. Stimuli were
presented on a computer monitor located 143 cm from
the subject. Images subtended an average of 4.8° (?
1.4°) of visual angle in the vertical plane and 4.4° (?
1.2°) in the horizontal plane.
Images were presented in accordance with the as-
cending method of limits (AML) (Fig. 1), from least
complete (Level 7) tomost complete (Level 1). After the
presentation of each fragmented image a “Y|N” cue
appeared, prompting a forced-choice response. Sub-
jects pressed one button for a “No” response, indicating
that they could not “close” and more information was
needed or a second button for a “Yes” response, indi-
cating that they could “close” and name the previous
fragmented image. Following “No” responses, subjects
were presented with the next most complete image of
the same picture and were again cued for a forced-
choice decision. Following “Yes” responses, the picture
sequencewas terminated and subjects wererequired to
verbally namethepicture. Theexperimenter then gave
a verbal “Go” cue, indicating that the subject should
press either button to initiate the next sequence of
The experiment consisted of 40 blocks, each block
containing 10 different picture sequences, of which 7
were presented only once and 3 were presented twice
(i.e., 13 picture sequences per block). Repeated picture
sequences consisted of theidentical fragmented images
as when initially presented. The positions of the to-be-
repeated picture sequences were randomly selected.
The number of picture sequences intervening between
initial and repeated presentations was either one or
two, determined at random. Subjects were encouraged
to take breaks between blocks whenever they deemed
it necessary to maintain high concentration and pre-
The timing of presentations (Fig. 1) was such that
each image appeared for 750 ms, followed by a blank
screen for 800 ms. Then a “Y|N” response prompt
appeared for 200 ms, followed by a blank screen for
2200 ms. Subjects’ response window extended for 2300
ms from the onset of the “Y|N” prompt. Use of the
response prompt was motivated by the desire todimin-
ish the impact of motor responses on the sensory ERP
to the pictures.
DONIGER ET AL.
Measurements and Analyses
High-density event-related potentials (ERP) were
acquired from 64 scalp electrodes referenced to nose
(band-filtered from 0.05 to100 Hz; digitized at 500 Hz;
impedances ? 5 k?). The basic layout is an extension
of the modified “International extended 10-20 system”
(American EEG Society, 1991) and can be seen in Fig.
2. Note that sites where analyses were conducted are
highlighted. Trials with blinks and large eye move-
ments were rejected off-line on the basis of horizontal
(HEOG) and vertical (VEOG) electro-oculogram (arti-
fact rejection window of ? 60 ?V). No systematic dif-
ferences in HEOG or VEOG were seen across condi-
tions. An artifact criteria of ? 60 ?V (N ? 8) or ? 70 ?V
(N ? 2) was used at all other electrode sites to reject
trials with excessive EMG or other noise transients.
Average rejection rates were 25.9% (? 11.3) with a
high of 48% for one subject and a low of 14% for the
least noisy subject. Accepted trials wereepoched (?100
ms prestimulus to 700 ms poststimulus) and then av-
eraged separately for each condition: ID, the level of
achieved during the first presentation of a picture se-
quence; 1-prior, thelevel of fragmentation immediately
preceding this ID level; Initial, the initial presentation
of a highly fragmented image that was not identified,
F IG. 1.
fragmented levels until identification is possible. The fourth picture sequence is a repeat of the first, but identification occurs at a more
fragmented level (Level 7), demonstrating the behavioral repetition priming effect. “Repeat” refers to an image at the level of identification
on the repeat picture sequence; “Initial” refers to the same image prior to identification on the initial presentation of the sequence. (b) The
most fragmented image (Level 7) onset at 0 ms (duration 750 ms), followed by a “Y|N” response prompt (R; duration 200 ms) at 1550 ms.
An “N” response resulted in presentation of the next level image, 2.2 s after the response prompt. A “Y” response terminated the sequence
for the subject’s verbal response. (c) Sample highly fragmented images (Level 6). For 10% of subjects, these images contained sufficient visual
information topermit object recognition when initially presented. After perception of the complete picture (see Fig. 7), 100% of subjects were
able to recognize the image on the left, and 90% could identify the remaining two images.
Stimulus configuration. (a) A series of four picture sequences from a typical block. Images are presented at successively less
VISUAL LEARNING IN HUMAN OBJ ECT RECOGNITION AREAS
but would subsequently be identified at this level when
repeated; Repeat, the level of fragmentation at which
object recognition was achieved for the repeat presen-
tation for which priming was demonstrated relative to
Initial (note that Repeat includes the identical highly
fragmented images as in Initial).
The average number of accepted sweeps per condi-
tion was: 304.4 (? 32.3) for ID; 300.1 (? 35.4) for
1-prior; 73.6 (? 15.8) for Initial; and 73.5 (? 16.2) for
Repeat. Baseline was defined as the mean voltage over
?100 to 20 ms. Scalp current density (SCD) topo-
graphic maps (second spatial derivative of the poten-
tial) werecomputed from spherical splineinterpolation
of the surface voltage recordings (see Perrin et al.,
1989), defining a quantity proportional to the magni-
tude of current-flow radial to the surface of the scalp.
SCD reduces the high degree of spatial overlap of the
ERPs that would otherwise be observed in the scalp
voltage data due to volume conduction, and eliminates
the effect of the reference electrode. Hence, SCD pro-
duces scalp topography maps that emphasize local
changes in ERP amplitude and are relatively insensi-
tive to contributions from remote generators.
Subjects correctly identified pictures 88% of the
time, with a modal level of identification (ID) for all
initially presented picture sequences of Level 3 (mean
proportion identified at Level 3 was 0.24), replicating
the findings of Doniger et al. (2000). Modal level of
identification for repeat picture sequences (Repeat)
was Level 7 (mean proportion identified at Level 7 was
0.41). The modal level of identification for the initial
presentations of this subset of picture sequences (Ini-
tial) was Level 3 (mean proportion identified at Level 3
was 0.27). Thus, subjects were often able to recognize
objects at their most fragmented level when these im-
ages had been previously presented. Subjects demon-
strated a substantial improvement (on theorder of four
levels of fragmentation) in their ability to achieve ob-
ject recognition from initial to repeat presentations of
the same picture sequences. Similarly, in examining
the cumulative proportion of pictures recognized by a
given level (Fig. 3), a significantly greater proportion of
pictures were recognized for repeat relative to initial
presentations for all but the two most complete levels.
There was no difference in the proportion recognizable
curves between initial presentations of pictures that
were subsequently repeated and presentations of pic-
tures that were not repeated.
F IG. 3.
formance curve versus level of fragmentation for initial and repeat
presentations of the same picture sequences.
Behavioral perceptual learning effect. Cumulative per-
F IG. 2.
event-related potentials (ERP) were recorded are represented on a
three-dimensional reconstruction from anatomic MRI (front and
back view). Sites used in analyses are highlighted.
Electode array. Positions of the 64 scalp sites from which
DONIGER ET AL.
F IG. 4.
maximal Nclamplitude, at the level of object identification (“ID”; black trace) and at the level prior to identification (“1-prior,” red trace). (b)
Group averaged (N ? 10) voltage waveforms at T5 and T6, for repeated images at the level of object identification (“Repeat”; blue trace) and
for the identical fragmented images when presented initially but not identifiable (“Initial”; green trace). (c) Scalp current density (SCD)
topographic maps (back views) of the difference waveform between level “ID” and “1-prior” at 275 ms (“ID”/“1-prior” peak) and 300 ms
(“Repeat”/“Initial” peak) poststimulus onset. Red isocontour lines (0.2 ?V/cm2) indicate positive values and blue, negative. A magenta disk
indicates the location of electrode T5 and a green disk the position of electrode T6. Bilateral negative foci, characteristic of Ncl, are evident
over occipitotemporal scalp. (d) SCD topographic maps of the difference waveform between “Repeat” and “Initial” presentations at 275 ms
and 300 ms poststimulus onset. Bilateral negative foci of the Ncleffect are again evident over occipitotemporal scalp.
Results. (a) Group averaged (N ? 10) voltage waveforms at left (T5) and right (T6) hemisphere occipitotemporal electrodes of
F IG. 5.
(blue trace) and “Initial” (green trace) presentations. An orange bar indicates a significant difference between conditions (P ? 0.05) for the
marked time-points. SCD topographic map (back view) of the difference waveform between “Repeat” and “Initial” presentations at 170 ms
poststimulus onset. Red isocontour lines (0.1 ?V/cm2) indicate positive values and blue, negative. Bilateral negative foci are evident over
N1 effect. Group averaged (N ? 10) voltage waveforms at the right hemisphere site of maximal N1 amplitude (P6), for “Repeat”
F IG. 6.
identification (“ID”; black trace) and at the level prior to identification (“1-prior,” red trace). Early stimulus-driven componentry is evident,
but there is no between-condition difference until after 300 ms, following peak Nclactivity.
Frontal scalp sites. Group averaged (N ? 10) voltage waveforms at left (F3) and right (F4) frontal sites, at the level of object
VISUAL LEARNING IN HUMAN OBJ ECT RECOGNITION AREAS
Group averaged visual evoked potentials (VEPs) for
the level ID and for the level prior to identification (1-
prior) replicated our findings (Doniger et al., 2000) of a
robust negative deflection in the 230 to 330 ms latency
range related to perceptual closure processes that we
have previously termed the Ncl(see Fig. 4a). A difference
in N1 amplitude was evident bilaterally between group
averaged VEPs for Repeat and Initial, but absent from
the ID/1-prior comparison (Figs. 4a and 4b). As in our
prior work, the large divergence that characterizes the
Nclarose just following the peak of the P2 component.
A pair of ANOVAs (2 conditions X 2 hemispheres X 3
electrodelocations) testedfor significant differencesin Ncl
amplitude between ID and 1-prior and between Repeat
and Initial. As SCD maps (Figs. 4c and 4d) interpolated
from all electrode sites confirmed that the topographic
distribution of Nclwas consistent with our prior study,
thefollowing pairs of sites weretested: P5/P6, T5/T6, and
PO5/PO6. Area measures were taken from a 20 ms win-
dow centered at peak Ncl latency for each comparison
(ID/1-prior ? 275 ms; Repeat/Initial ? 300 ms). The ID/
1-prior ANOVA revealed a significant main effect of con-
dition (F(1,9) ? 29.06, p ? 0.0001) and a significant
condition X hemisphere interaction (F(1,9) ? 5.94, p ?
0.04), reflecting greater amplitude at ID over the right
hemisphere. The ANOVA that tested for Ncl enhance-
ment between Initial and Repeat presentations of the
samefragmented imagealsoproduced a significant main
effect of condition (F(1,9) ? 40.96, p ? 0.0001) and a
significant condition X hemisphere interaction (F(1,9) ?
6.85, p ? 0.03).
Nclfoci for theRepeat/Initial comparison havehigher
amplitude than those for the ID/1-prior comparison.
Recall that subjects typically achieved object recogni-
tion after five levels of fragmentation had been pre-
sented. Further, consistent with our prior work, recall
that presentation of successively more complete levels
was accompanied by an incremental increase in Ncl
amplitude (Doniger et al., 2000). Also recall that for
repeat picture sequence presentations, subjects identi-
fied objects most often at their most fragmented level.
Hence, images in the Initial condition were presented
approximately four levels of fragmentation prior to
object recognition, whereas the identical images in the
Repeat condition were presented at object recognition.
This likely accounts for the greater amplitude of Ncl
foci in the Repeat/Initial comparison (Fig. 4c) relative
to the ID/1-prior comparison (Fig. 4d).
Wetested for point of Nclonset with a series of paired
two-tailed t tests between ID and 1-prior at the three
representativepairs of electrodesites used in theabove
analyses (P5/P6, T5/T6, and PO5/PO6). Tests werecon-
ducted at latencies preceding the Nclpeak to mark the
earliest timepoint that conformed to a 0.05 criterion.
Onset latencies across the three left and across the
three right hemisphere electrode sites were averaged
to provide a best estimate of Nclonset in a given hemi-
sphere. A point was only accepted as the earliest diver-
gence if at least 11 subsequent consecutive time-points
(?20 ms at 500 Hz digitization rate) met the 0.05
criterion (see Guthrie and Buchwald, 1991). The crite-
rion was met at 236 ms for the right hemisphere and
241 ms for theleft. Theseonset latencies areconsistent
with thoseof our previous study whereNclwas found to
onset at 232 ms in both hemispheres (Doniger et al.,
2000). As Nclonsets just after the peak of the P2 com-
ponent, the question arises whether P2 and Ncl are
both modulated by perceptual closureprocesses. In this
regard, we found peak P2 amplitude tobe at 229 ms at
electrodesiteP3. That maximal P2 amplitudepreceded
Nclonset and that it peaks at a more medial electrode
sitesuggests that P2 generators aredistinct from those
that give rise to Ncl. We found no significant modula-
tion of the P2 component at any scalp site.
Another pair of ANOVAs (2 conditions X 2 hemi-
spheres X 3 electrode locations) was conducted to exam-
ine differences in N1 amplitude in the ID/1-prior and
Repeat/Initial comparisons. Area measures were taken
froma 10-ms window (electrodes: P5/P6, T5/T6, andPO5/
PO6) centered at the peak latency for N1 in each com-
parison (ID/1-prior ? 165 ms; Repeat/Initial ? 175 ms).
The Repeat/Initial ANOVA yielded a significant main
effect of condition (F(1,9) ? 6.66, P ? 0.04), which was
absent from the ID/1-prior ANOVA (F(1,9) ? 0.08, P ?
0.78). Further, post-hoc analyses revealed that the mag-
nitude of the difference between conditions for the Re-
peat/Initial comparison was significantly greater than
that for the ID/1-prior comparison (F(1,9) ? 5.48, P ?
0.05). Both the Repeat/Initial and ID/1-prior ANOVAs
yielded significant main effects of hemisphere (Repeat/
Initial: F(1,9) ? 10.46, P ? 0.02; ID/1-prior: F(1,9) ? 5.78,
P ? 0.05). No other significant effects were seen. Fol-
low-up planned comparisons (2-tailed t tests) showed the
N1 effect tobe highly robust at the scalp site of maximal
effect (site P6) (t9 ? 3.80, P ? 0.005). The location of
neural generators contributing to the N1 effect was esti-
mated with an SCD map generated from the difference
wave between the ERP to Repeat and Initial (Fig. 5). At
the peak of this N1 difference, bilateral occipitotemporal
foci, similar to those of the subsequent Ncl, were appar-
In order to support our hypothesis that the N1 rep-
etition effect results from priming-related identifica-
tion, we conducted a control analysis in which we ex-
amined N1 amplitude for
remained unidentifiablewhen repeated (?59% of Level
7 presentations were not identified upon repeat pre-
sentation). Nonidentified images at Level 7 on the re-
peat presentation of the picture sequence (RNID) were
compared to the same nonidentified images on the
initial picture sequence presentation (INID). The
RNID/INID ANOVA (2 conditions X 2 hemispheres X 3
electrode locations) showed no significant modulation
Level 7 images that
DONIGER ET AL.
of N1 (F(1,9) ? 1.26, P ? 0.29). There were no other
significant main effects or interactions.
Wetested for point of N1 onset with a series of paired
two-tailed t tests between Initial and Repeat at the
electrode sites used in the above analyses. The site of
maximal N1 difference was at site P6, which showed
an onset conforming tothe11 point criterion at 132 ms.
None of the differences at left-hemisphere sites con-
formed to this criterion.
We also investigated whether conditional effects
were seen over frontal sites that either preceded or
were contemporaneous with the effects over lateral
occipital sites. Multiple 2-tailed paired t test compari-
sons in the period from 0 to 300 ms yielded no signifi-
cant differences over any frontal scalp sites (seeFig. 6).
Late differences after 300 ms and after the peak of Ncl
were observed and will be the subject of future inves-
tigations. Note that early stimulus-driven componen-
try is evident over these frontal scalp sites (Fig. 6) but
is not modulated by perceptual closure processes.
Thecurrent findings elucidatethebrain mechanisms
that subserve visual perceptual learning in humans.
We define an electrophysiological correlate of repeti-
tion priming due to perceptual learning over LO com-
plex that precedes closure processes over the same
region. Localization of priming effects to the LO com-
plex is consistent with findings from hemodynamic im-
aging studies in humans (e.g., Dolan et al., 1997; J ames
et al., 1999; Badgaiyan et al., 1999; Buckner et al.,
2000; Henson et al., 2000) and single-unit studies in
the corresponding region (area IT) of macaque visual
cortex (e.g., Tovee et al., 1996). Brain operations re-
lated torepetition priming in the present study appear
to differ from those related to perceptual closure, in
that, priming produces enhancement of both the N1
and Nclcomponents of the ERP, while closure effects
are confined to the latter component. That Ncl-en-
hancement is observed to the same highly fragmented
image after perceptual learning has occurred indicates
that Nclis largely dissociable from the physical param-
eters of the visual stimulation itself, but rather, in-
dexes perceptual closure and object recognition pro-
cesses (see also Doniger et al., 2000).
Recognition of a highly impoverished stimulus dueto
perceptual learning requires access toa representation
or sensory trace laid down when “closure” was previ-
ously achieved. A central question in repetition prim-
ing studies is the neural locus of this trace. Candidate
regions include posterior object recognition areas (e.g.,
Lueschow et al., 1994; Desimone, 1996; Gibson and
Maunsell, 1997; Chelazzi et al., 1998; but see Miller et
al., 1993), perirhinal cortex (see Sakai and Miyashita,
1991; Murray and Bussey, 1999), and prefrontal work-
ing memory areas (e.g., Miller et al., 1993; Levy and
Goldman-Rakic, 1999; J iang et al., 2000), all of which
show delay activity during explicit memory tasks. It is
important to point out that these studies showing
memory-related delay activity employ distinctly differ-
ent paradigms to the one used in the current study. In
particular, the delay periods are usually far shorter
than the period between the Initial and Repeat presen-
tation in our study and do not involve the interim
presentation of many other stimuli. Nonetheless, these
studies implicate the above mentioned areas as likely
candidates for maintenance of a perceptual trace.
While not disputing the obvious possibility that main-
tenance of such a trace is shared between inferotem-
poral and prefrontal areas, our finding that theearliest
priming-related differentiation is an N1-enhancement
over LO suggests that these object recognition areas
may play a predominant role in trace maintenance.
That we do not find such early differences over frontal
scalp is consistent with the idea that the memory pro-
cesses associated with priming may not involvea work-
ing memory component, normally associated with fron-
tal areas (see Fig. 6).
Another key finding of the current study is that
while both hemispheres are activated during percep-
tual closure processes (also Doniger et al., 2000), right
hemisphere activation is of significantly greater ampli-
tude than left. A prominent hypothesis holds that the
right hemisphere is more involved in the processing of
global form, while the left hemisphere is more involved
in analytic local processing (see e.g., Robertson et al.,
1988; Atchley and Atchley, 1998; Corballis et al., 1999).
In this vein, the greater right-hemisphere activation in
this study might reflect a relatively greater contribu-
tion of global processing toperceiving a complete object
from fragmentary evidence. This view is supported by
both electrophysiological and hemodynamic studies
whereselectiveattention toglobal versus local features
of a stimulus array was manipulated (e.g., Heinze et
al., 1998; Fink et al., 1997), and by neuropsychological
studies of perceptual closure and amodal completion
(e.g., Wasserstein et al., 1984). Critically, despite the
asymmetry in activation amplitudes across hemi-
spheres, we find that Ncl onsets at the same time in
both the right and left hemispheres, suggesting that
both hemispheres access object identity information in
The N1 effect is ?100 ms earlier than Ncl, yet has a
very similar scalp-topography. This suggests that the
N1 effect and Nclmay reflect successive stages of dif-
ferential activation in areas of the LO complex. This
F IG. 7. Complete versions of the images shown in Fig. 1c.
VISUAL LEARNING IN HUMAN OBJ ECT RECOGNITION AREAS
progression of events can be interpreted within a the-
oretical framework that posits a distinction between
“perceptual” and “conceptual” modes of object recogni-
tion (see Tulving and Schacter, 1990; Schacter, 1992).
That is, the “perceptual mode” would describe repeti-
tion priming due to perceptual learning, where object
recognition is facilitated by a presemantic (sensory)
trace laid down by recent sensory exposure to an item,
and is rapid and effortless. In contrast, the “conceptual
mode” would describe object recognition during initial
picturesequencepresentations, as sensory information
is actively compared with semantic/episodic memory
representations. Within this theoretical framework,
the N1 may represent structural analysis of the object
(i.e., perceptual mode) that is enhanced for Repeat
versus Initial presentations, reflecting access to the
sensory trace laid down when closure was first
achieved. In turn, the later Nclcomponent (?290 ms)
might be biased towards semantic processing (i.e., con-
ceptual or memory mode). Indeed Ncl appears only
when effort is required to decipher incomplete images,
whereas when complete and easily-identified objects
are presented, only N1 is modulated. N1 activity in the
vicinity of LO has been found in studies of face-, object-,
and word-recognition (e.g., Allison et al., 1999; Roission
et al., 2000; Bentin et al., 1999). Moreover, it has been
suggested that this activity reflects the structural en-
coding of object components and may be particularly
sensitive to the presence of particular object compo-
nents (Eimer, 1998; Bentin et al., 1996). Finally, it
should be noted that structural processing, as reflected
by N1, might facilitate the subsequent conceptual pro-
cessing reflected by Ncl. Hence, the N1/Ncl sequence
may constitute evidence for an important object-pro-
cessing system within LO.
It is important to note that while we interpret the
N1- and Ncl- enhancements as reflecting successive
stages of object-processing within the LO complex
based on their very similar scalp topographies (Figs.
4c, 4d, and 5), this is not to imply that only a single
generator is responsible for both effects. In fact, while
the SCD topographic maps provide compelling evi-
dence for generators in inferotemporal cortices, it is
also evident that both modulations (N1 and Ncl) repre-
sent contributions from morethan a singleintracranial
generator. It is likely that the LO complex comprises a
number of distinct functional units (e.g., Ishai et al.,
1999), which may play differential roles in priming-
related versus closure-related processes. Indeed, intra-
cranial recordings in humans have shown at least two
distinct areas of the occipitotemporal cortex that con-
tribute to object recognition (Allison et al., 1999). Ad-
ditionally, our effects are likely to also represent some
contribution from lower-tier extrastriate regions as the
complex operations of perceptual closure likely involve
recursive feedback-feedforward mechanisms between
areas at different levels of the cortical hierarchy (e.g.,
Pollen, 1999; see Doniger et al., 2000, for a more de-
tailed treatment). Future studies combining the high
spatial resolution of hemodynamic imaging with the
fine temporal resolution that can only be provided by
high-density electrical mapping, should allow for fur-
ther dissociation of these functional units and elucida-
tion of their temporal relations (e.g., Simpson et al.,
1995; Martinez et al., 1999).
In conclusion, the current study extends previous
work in finding a sequence of electrical events that
indexes repetition priming due to visual perceptual
learning in humans and in detailing the timecourse of
these events. We provide evidence for a neural corre-
late of repetition priming as evidenced by enhance-
ment of the N1 component of the ERP followed by
closure-related Ncl-enhancement. The N1/Nclsequence
may reflect successive stages of object processing in
regions of the LO complex, with N1-enhancement re-
flecting moreperceptual structural analysis of thefrag-
mented images and Nclreflecting more conceptual ef-
fortful analysis. Within this framework, our data
underscore the importance of considering the temporal
dimensions of the visual processing hierarchy (e.g.,
Nowak and Bullier, 1997; Schroeder et al., 1998).
Support in part from NIMH (MH49334, MH01439). Sincere appre-
ciation to Dr. J in Fan for invaluable programming assistance and to
Drs. Walter Ritter and J oan Gay Snodgrass for insightful discus-
sions. We also thank Mr. Paul Higgins for technical assistance.
Allison, T., Puce, A., Spencer, D., and McCarthy, G. 1999. Electro-
physiological studies of human face perception I: Potentials gen-
erated in occipitotemporal cortex by face and non-face stimuli.
Cereb. Cortex 9:415–430.
Atchley, R. A., and Atchley, P. 1998. Hemispheric specialization in
the detectionof subjective
Badgaiyan, R. D., Schacter, D. L., and Alpert, N. M. 1999. Auditory
priming within and across modalities: Evidence from positron
emission tomography. J . Cogn. Neurosci. 11:337–348.
Bartlett, F. C. 1916. An experimental study of some problems of
perceiving and imagining. Brit. J . Psychol. 8:222–266.
Bentin, S., Allison, T., Puce, A., Perez, E., and McCarthy, G. 1996.
Electrophysiological studies of face perception in humans. J . Cogn.
Bentin, S., Mouchetant-Rostaing, Y., Giard, M. H., Echallier, J . F.,
and Pernier, J . 1999. ERP manifestations of processing printed
words at different psycholinguistic levels: Time course and scalp
distribution. J . Cogn. Neurosci. 11:235–260.
Buckner, R. L., Koutstaal, W., Schacter, D. L., Rosen, B. R. 2000.
Functional MRI evidence for a role of frontal and inferior temporal
cortex in amodal components of priming. Brain 123:620–640.
Chelazzi, L., Duncan, J ., Miller, E. K., and Desimone, R. 1998.
Responses of neurons in inferior temporal cortex during memory-
guided visual search. J . Neurophysiol. 80:2918–2940.
Corballis, P. M., Fendrich, R., Shapley, R. M., and Gazzaniga, M. S.
1999. Illusory contour perception and amodal boundary comple-
tion: Evidence of a dissociation following callosotomy. J . Cogn.
DONIGER ET AL.
Cycowicz, Y. M., Friedman, D., Rothstein, M., and Snodgrass, J . G.
1997. Picture naming by young children: Norms for name agree-
ment, familiarity, and visual complexity. J . Exp. Child Psychol.
Desimone, R. 1996. Neural mechanisms for visual memory and their
role in attention. Proc. Natl. Acad. Sci. USA 93:13494–13499.
Dolan, R. J ., Fink, G. R., Rolls, E., Booth, M., Holmes, A., Frackow-
iak, R. S., and Friston. K. J . 1997. How the brain learns to see
objects and faces in an impoverished context. Nature389:596–599.
Doniger, G. M., Foxe, J . J ., Murray, M. M., Higgins, B. A., Snodgrass,
J . G., Schroeder, C. E., and J avitt, D. C. 2000. Activation time-
course of ventral visual stream object-recognition areas: High den-
sity electrical mapping of perceptual closure processes. J . Cogn.
Eimer, M. 1998. Does the face-specific N170 component reflect the
activity of a specialized eye processor? Neuroreport 9:2945–2948.
Fink, G. R., Halligan, P. W., Marshall, J . C., Frith, C. D., Frackow-
iak, R. S., and Dolan, R. J . 1997. Neural mechanisms involved in
the processing of global and local aspects of hierarchically orga-
nized visual stimuli. Brain 120:1779–1791.
Foley, M. A., Foley, H. J ., Durso, F. T., and Smith, N. K. 1997.
Investigations of closure processes: What source-monitoring judg-
ments suggest about what is “closing.” Mem. Cognit. 25:140–155.
Gibson, J . R., and Maunsell, J . H. 1997. Sensory modality specificity
of neural activity related to memory in visual cortex. J . Neuro-
Grill-Spector, K., Kushnir, T., Edelman, S., Avidan, G., Itzchak, Y.,
and Malach, R. 1999. Differential processing of objects under var-
ious viewing conditions in the human lateral occipital complex.
Guthrie, D., and Buchwald, J . S. 1991. Significance testing of differ-
ence potentials. Psychophysiology 28:240–244.
Haxby, J . V., Ungerleider, L. G., Clark, V. P., Schouten, J . L., Hoff-
man, E. A., and Martin, A. 1999. The effect of face inversion on
activity in human neural systems for face and object perception.
Heinze, H. J ., Hinrichs, H., Scholz, M., Burchert, W., and Mangun,
G. R. 1998. Neural mechanisms of global and local processing: A
combined PET and ERP study. J . Cogn. Neurosci. 10:485–498.
Henson, R., Shallice, T., andDolan, R. 2000. Neuroimaging evidencefor
dissociable forms of repetition priming. Science 287:1269–1272.
Ishai, A., Ungerleider, L. G., Martin, A., Schouten, J . L., and Haxby,
J . V. 1999. Distributed representation of objects in the human
ventral visual pathway. Proc. Natl. Acad. Sci. U S A 96:9379–
Ito, M., Tamura, H., Fujita, I., and Tanaka, K. 1995. Size and
position invariance of neuronal responses in monkey inferotempo-
ral cortex. J . Neurophysiol. 73:218–226.
J ames, T. W., Humphrey, G. K., Gati, J . S., Menon, R. S., and
Goodale, M. A. 1999. Repetition priming and the time course of
object recognition: An fMRI study. Neuroreport 10:1019–1023.
J iang, Y., Haxby, J . V., Martin, A., Ungerleider, L. G., and Parasura-
man, R. 2000. Complementary neural mechanisms for tracking
items in human working memory. Science 287:643–646.
Kanwisher, N., Woods, R. P., Iacoboni, M., and Mazziotta, J . C. 1997.
A locus in human extrastriate cortex for visual shape analysis. J .
Cognit. Neurosci. 9:133–142.
Levy, R., and Goldman-Rakic, P. S. 1999. Association of storage and
processing functions in the dorsolateral prefrontal cortex of the
nonhuman primate. J . Neurosci. 19:5149–5158.
Lueschow, A., Miller, E. K., and Desimone, R. 1994. Inferior tempo-
ral mechanisms for invariant object recognition. Cereb. Cortex
Malach, R., Reppas, J . B., Benson, R. R., Kwong, K. K., J iang, H.,
Kennedy, W. A., Ledden, P. J ., Brady, T. J ., Rosen, B. R., Tootell,
R. B. 1995. Object-related activity revealed by functional magnetic
resonance imaging in human occipital cortex. Proc. Natl. Acad.
Sci. USA 92:8135–8139.
Martinez, A., Anllo-Vento, L., Sereno, M. I., Frank, L. R., Buxton,
R. B., Dubowitz, D. J ., Wong, E. C., Hinrichs, H., Heinze, H. J ., and
Hillyard, S. A. 1999. Involvement of striate and extrastriate visual
cortical areas in spatial attention. Nat. Neurosci. 2:364–369.
Miller, E. K., Li, L., and Desimone, R. 1993. Activity of neurons in
anterior inferior temporal cortex during a short-term memory
task. J . Neurosci. 13:1460–1478.
Murray, E. A., and Bussey, T. J . 1999. Perceptual-mnemonic func-
tions of the perirhinal cortex. Trends Cogn. Sci. 3:142–151.
Nowak, L. G., and Bullier, J . 1997. The timing of information trans-
fer in the visual system. In Cerebral Cortex, Vol. 12, Extrastriate
Cortex (K. Rockland, J . Kaas, and A. Peters, Eds.), pp. 205–241.
Plenum Press, New York.
Perrin, F., Pernier, J ., Bertrand, O., Echallier, J . F. 1989. Spherical
splines for scalp potential and current density mapping. Electro-
encephalogr. Clin. Neurophysiol. 72:184–187.
Pollen, D. A. 1999. On the neural correlates of visual perception.
Cereb. Cortex 9:4–19.
Robertson, L. C., Lamb, M. R., and Knight, R. T. 1988. Effects of
lesions of temporal-parietal junction on perceptual and attentional
processing in humans. J . Neurosci. 8:3757–3769.
Roission, B., Gauthier, I., Tarr, M. J ., Despland, P., Bruyer, R.,
Linotte, S., and Crommelinck, M. 2000. The N170 occipito-tempo-
ral component is delayed and enhanced toinverted faces but not to
inverted objects: An electrophysiologic account of face-specific pro-
cesses in the human brain. Neuroreport 11:69–74.
Sakai, K., and Miyashita, Y. 1991. Neural organization for the long-
term memory of paired associates. Nature 354:152–155.
Sary, G., Vogels, R., and Orban, G. A. 1993. Cue-invariant shape selec-
tivity of macaque inferior temporal neurons. Science 260:995–997.
Schacter, D. L. 1992. Priming and multiple memory systems: Percep-
tual mechanisms of implicit memory. J . Cogn. Neurosci. 4:244–256.
Schroeder, C. E., Mehta, A. D., and Givre, S. J . 1998. A spatiotem-
poral profile of visual system activation revealed by current source
density analysis in the awake macaque. Cereb. Cortex 8:575–592.
Simpson, G. V., Pflieger, M. E., Foxe, J . J ., Ahlfors, S. P., Vaughan,
H. G. J r., Hrabe, J ., Ilmoniemi, R. J ., and Lantos, G. 1995. Dy-
namic neuroimaging of brain function. J . Clin. Neurophys. 12:432–
Snodgrass, J . G., and Corwin, J . 1988. Perceptual identification
thresholds for 150 fragmented pictures from the Snodgrass and
Vanderwart picture set. Percept. Mot. Skills 67:3–36.
Snodgrass, J . G., and Feenan, K. 1990. Priming effects in picture
fragment completion: Support for the perceptual closure hypothe-
sis. J . Exp. Psychol. Gen. 119:276–296.
Snodgrass, J . G., and Vanderwart, M. 1980. A standardized set of
260 pictures: Norms for name agreement, image agreement, famil-
iarity, and visual complexity. J . Exp. Psychol. Hum. Learn. Mem.
Tovee, M. J ., Rolls, E. T., and Ramachandran, V. S. 1996. Rapid
visual learning in neurones of the primate temporal visual cortex.
Tulving, E., and Schacter, D. L. 1990. Priming and human memory
systems. Science 247:301–306.
Vogels, R. 1999. Categorization of complex visual images by rhesus
monkeys. Part 2: Single-cell study. Eur. J . Neurosci. 11:1239–1255.
Wasserstein, J ., Zappulla, R., Rosen, J ., and Gerstman, L. 1984.
Evidence for differentiation of right hemisphere visual perceptual
functions. Brain Cogn. 3:51–56.
VISUAL LEARNING IN HUMAN OBJ ECT RECOGNITION AREAS