Boehm, S.G., Sankar, S.K., Klostermann, E.C., Florczak, S.M., & Paller, K.A. (2004, April). Recognition of verbal stimuli is driven by facilitation in both perceptual processing and memory access.

Florczak, S.M., Boehm, S.G., Rao, S.M., & Paller, K.A. (2004, April). Neural underpinnings of face familiarity and face recollection.

Li, W., Paller, K.A., & Zinbarg, R.E. (2004, April). Emotional stroop interference for subliminal and supraliminal word presentations: Behavioral and electrophysiological evidence.

Muthu, K., Kim, Y.-J., Grabowecky, M., Paller, K.A., & Suzuki, S. (2004, April). Steady-state visual evoked potentials (SSVEPs) and manipulations of stimulus contrast, flicker frequency, and selective attention.

Ken A. Paller and Craig A. Hutson. Department of Psychology and Institute for Neuroscience, Northwestern University, Evanston, IL, USA.

Measuring neural processes responsible for different types of memory is difficult because they tend to occur concurrently. Previously we isolated brain-potential correlates of face recollection using memory dissociations (Paller et al., 1999, 2000). Here, we attempted to disentangle neural events that support priming versus recognition by producing priming in the absence of recognition. Some faces were presented centrally for 105 ms, followed by a mask, while subjects made difficult visual discriminations on a parafoveal cross in a divided-attention task adapted from Smith and Merikle (2000). Other faces, which subjects tried to remember, were presented for 300 ms without a mask or cross. Thesestudied-brief and studied-long faces were presented again ~2 min later while subjects discriminated between faces viewed earlier and new faces. Memory results showed that studied-brief faces were primed (speeded fame judgments) but not remembered (chance recognition levels), whereas studied-long faces were well-remembered. Electrophysiological results revealed (1) a negative brain potential (250-400 ms, right-central and frontal) that we postulate to be a neural correlate of processing that supports priming for faces but not episodic recognition, and (2) a positive brain potential (350-900 ms, posterior topography) that we take as a neural correlate of face recollection. Absence of recollection may be necessary for observing this neural correlate of face priming, given that simultaneous indicators of priming and recollection can cancel out. Characterizing neural activity associated with priming and recollection can thus contribute to elucidating the biological bases of distinctions between conscious and nonconscious memory. (Support: NINDS)

READING THE MIND'S EYE USING AN OCCIPITAL BRAIN POTENTIAL THAT INDEXES VIVIDNESS OF VISUAL IMAGERY.

Brian Gonsalves, Paul J. Reber, Malena Crawford, & Ken A. Paller. Department of Psychology, Northwestern University, Evanston, IL.

Previously we found that an occipital brain potential, labeled P850, was larger when people generated visual images in response to spoken words, compared to when they attended to the sounds and letters in each word (Gonsalves & Paller, 2000a, Memory & Cognition). P850 potentials were also observed during memory testing and were found to be enhanced when people accurately retrieved visual images. Furthermore, P850 amplitude was significantly greater when robust visual image generation led to subsequent false remembering (Gonsalves & Paller, 2000b, Nature Neuroscience). Here, we tested the hypothesis that P850 amplitude varies as a function of the subjective vividness of visual imagery. We presented a series of spoken nouns and asked participants to attempt to mentally generate a vivid visual image of the referent of each word. Participants then rated imagery vividness on a four-point scale, trying to use each category equally often. Potentials recorded during image generation were computed as a function of vividness ratings. Results showed that P850 amplitude increased as vividness ratings increased. Our findings suggest that this brain potential reflects processes associated with generating visual images. Furthermore, fMRI results have implicated neural activity in the precuneus region during visual image generation, and in particular when that imagery leads to the subsequent tendency to mistakenly attribute memory for that imagery to a memory for a perceptual experience (Gonsalves et al., 2001, SFN-abstracts). We thus speculate that P850 potentials and activation of the precuneus are associated with bringing a stored representation of a visual object to mind.

DOES THE BRAIN EQUAL THE SUM OF ITS HEMISPHERES? INTERHEMISPHERIC COMMUNICATION IN FACE PERCEPTION.

Galit Yovel1 Ken A. Paller2 & Jerre Levy1
1University of Chicago 2Northwestern University

The left and right hemispheres use complementary strategies in processing upright faces, but similar strategies in processing inverted faces. In the current study, we examine the effects of interhemispheric integration on face perception when the two hemispheres process consistent or inconsistent input in a complementary (upright faces) or in a similar (inverted faces) manner. Participants performed a face matching task in which four types of facial stimuli were presented foveally in an upright or inverted orientation: Bisymmetric (B) faces (bilaterally consistent), chimeric (C) faces (bilaterally inconsistent), and two types of hemifaces, in which only a left (L) or right (R) half (defined by the viewerÕs perspective) presents useful facial information. If the two hemispheres process information independently, the accuracy for B-faces and C-faces would equal the union of accuracies for L-faces and R-faces. Performance was better for B-faces (i.e., facilitation) and worse for C-faces (i.e., interference) than the predicted performance of two independent hemispheres. Facilitation was larger than interference for both face types and facilitation and interference were larger for upright than inverted faces. The larger facilitation for upright than inverted B-faces reflects the advantage of interhemispheric integration when the two hemispheres make complementary contributions to information processing. The larger interference for upright than inverted C-faces together with the larger facilitation than interference for both face types suggest that interhemispheric integration is interrupted when the two hemispheres process inconsistent information, and that this occurs earlier when processing is similar (inverted C-faces) than when it is complementary (upright C-faces).

NEURAL CORRELATES OF IMPLICIT MEMORY FOR COMMON OBJECTS.

Stephan G. Boehm, Brian Gonsalves, Ellen C. Klostermann & Ken A. Paller, Department of Psychology, Northwestern University, Evanston, IL.

Human memory has been investigated using event-related potentials (ERPs) to show that brain responses to stimuli are altered as a result of prior exposure to those specific stimuli, usually words or faces. Here, we attempted to broaden the scope of these investigations by measuring ERPs associated with implicit memory for color photographs of common objects. However, putative neural correlates of implicit memory can be contaminated by concurrent activity associated with recognition. Our approach to this problem was to record responses to repeated nontarget events embedded in a stimulus sequence which people monitored for the occurrence of target stimuli (a modified Sternberg task), on the assumption that recognition processes for repeated nontargets will be minimized in this situation. Previously, Boehm, Lueschow and Sommer (2000) used a related experimental design with word and face stimuli, and showed that ERP differences between repeated and new nontargets were unaffected by an encoding manipulation and were therefore interpreted as ERP correlates of implicit memory. Using this same experimental design with objects, we observed positive ERP differences between repeated and new nontarget objects from approximately 150-400 ms, which we interpret as neural correlates of implicit memory. ERPs to target events included a positive deflection in the same latency range but with a much higher amplitude. Our results provide a way to investigate changes in the neural processing of visual objects that result from prior experience and that may support implicit memory.

Society for Neuroscience 2001

EVENT-RELATED fMRI REVEALS BRAIN ACTIVITY AT ENCODING THAT PREDICTS TRUE AND FALSE MEMORY FOR VISUAL OBJECTS.

B. Gonsalves*, P.J. Reber, D.R. Gitelman, T.B. Parrish, M.-M. Mesulam, and K.A. Paller, Depts. of Psychology, Neurology, and Radiology, Northwestern University, Evanston IL 60208.

One way to investigate memory storage is to analyze brain activity at the time of encoding as a function of the success of subsequent remembering. Previously we showed that event-related brain potentials at 300-1800 ms were larger for pictures remembered later, and that posterior potentials at 600-900 ms were larger for words falsely remembered as pictures (Gonsalves & Paller, Nature Neuroscience, 2000). We interpreted the posterior potentials as indications of visual imagery generated in response to words. Here we measured event-related fMRI responses in a similar paradigm. Subjects read names of objects, half of which were followed 2 s later by a picture of the object. Then, outside the scanner, subjects listened to object names and decided whether they had seen a picture of the corresponding object. Subjects mistakenly reported seeing some objects that they had not seen but rather had imagined in the encoding phase, when cued with the corresponding word. For encoding trials associated with these false memories, words elicited activation in the precuneus, left inferior occipital gyrus, and anterior cingulate. The left middle frontal gyrus was more active for words that did not lead to false memories. A different set of brain areas was predictive of accurate memory for pictures, including left inferior frontal gyrus, parahippocampal gyrus, and fusiform gyrus. The brain potential and fMRI evidence together support the idea that false memories can arise as a consequence of the activation of a brain network associated with visual imagery at the time of encoding. (Support: NINDS)

BRAIN NETWORKS FOR GAZE PROCESSING REVEALED WITH FMRI.

C.I. Hooker*, K.A. Paller, D.R. Gitelman, T.B. Parrish, M.-M. Mesulam, & P.J. Reber, Depts. of Psychology, Neurology, and Radiology, Northwestern Univ., Evanston IL 60208.

Using gaze direction to determine where someone is looking is important for social interaction. Prior neuroimaging suggests that the superior temporal sulcus (STS) is active when subjects view gaze cues, repetitive eye movements, and biological motion. We investigated the brain networks responsible for gaze perception using a task that required subjects to make precise discriminations concerning direction of gaze. In control tasks an arrow provided directional information instead of the eyes, or the eyes moved without providing relevant directional information. Whole-brain fMRI results (N=10) revealed greater STS activation when directional cues came from gaze than from the arrow, and greater STS activation from eye motion when it provided directional information than when it didn't. Activity in fusiform face-responsive areas and prefrontal cortex was greater in the eye-motion control task compared to the gaze task, and greater when attending to gaze than when attending to the arrow, even when the arrow was superimposed on the face. These results help define the interacting networks mediating face and gaze perception, and suggest that processing in the STS is instrumental for analyzing meaningful eye motion. In a second task, we investigated responses to gaze directed towards the subject, who was required to detect direct gaze trials among averted gaze trials. In a blocked design, direct gaze occurred on 40% or 0% of the trials, and amygdala activity was greater in the latter condition, when the unfulfilled anticipation of direct gaze was most prominent. These data support the role of the amygdala in monitoring emotionally relevant events.


Cognitive Neuroscience Society 2001

AN ERP ANALYSIS OF FACE MEMORY AIMED AT ISOLATINGPERCEPTUAL PRIMING FROM RECOGNITION EFFECTS.

Ken A. Paller, Brian Gonsalves, & Craig A. Hutson. Department of Psychology, Northwestern University, Evanston, IL.

When event-related potentials (ERPs) are compared for repeated versus nonrepeated stimuli, resulting ERP repetition effects include effects related to both explicit and implicit memory performance. Substantial evidence suggests that a portion of the differential brain activity produced by repetition reflects recognition memory. On the other hand, less research has focused on whether ERPs are specifically sensitive to processing that leads to perceptual priming (a distinct type of memory preserved in amnesic patients with impaired recognition). Nonetheless, prior experiments using study-format manipulations have shown that posterior ERPs at approximately 300-500 ms are associated with priming of visual word form (Paller et al., 1998ab). Here we attempted to isolate ERPs associated with face priming. At study, faces were viewed centrally for a brief duration and followed by a mask, and attention was directed to parafoveal cross stimuli presented simultaneously. This divided-attention procedure greatly reduced recognition memory but there was still an influence on how faces were processed when presented again 1-2 min later. Primary ERP differences took the form of enhanced centroparietal positivity around 400-500 ms. These results suggest that altered face processing in this paradigm differs from effects produced when faces are explicitly remembered, as in our prior studies of ERP correlates of face recollection (Paller et al., 1999, 2000), supporting current views on multiple memory systems.

AN FMRI ANALYSIS OF GAZE PROCESSING.

Christine I. Hooker, Ken A. Paller, Darren R. Gitelman, Todd B. Parrish, M.-Marsel Mesulam, & Paul J. Reber. Departments of Psychology, Neurology, and Radiology, Northwestern University.

Using direction of gaze to determine where another person is attending is an important aspect of social interaction. Previous neuroimaging results suggest that the superior temporal sulcus (STS) is active when attending to gaze direction, eye movements, and biological motion (Allison et al., 2000, TICS). We attempted to define more specifically the brain networks responsible for gaze perception by using a novel task that required subjects to make precise discriminations concerning direction of gaze. In control tasks an arrow provided directional information instead of the eyes, or the eyes moved without providing relevant directional information. In a group of 10 participants, STS was activated more by direction cues from gaze than from the arrow, and more by eye motion when it provided directional information than when it didn't. These data support the idea that processing in STS contributes to the analysis of meaningful eye motion that can then influence direction of attention. Furthermore, activity in fusiform face-responsive areas was greater when attending to gaze compared to the arrow on a face, and greater in the eye-motion control task compared to the gaze task. These results combined with other activations in parietal, frontal, and occipital cortices for gaze help define the interacting networks that mediate face and gaze perception, and may have implications for understanding pathological dysfunctions in these processes.

ELECTROPHYSIOLOGICAL CORRELATES OF PROCESSING STAGES IN FACE PERCEPTION.

Galit Yovel*, Jerre Levy*, Marcia Grabowecky** and Ken A. Paller**. *University of Chicago, Chicago, IL, **Northwestern University, Evanston, IL.

The present study investigated hemispheric differences and face-processing stages. Event-related potentials (ERPs) were recorded during a face-matching task for four types of centrally presented faces: Bisymmetric faces, comprised of a person's hemiface and its mirror image; Chimeric faces, which joined the hemifaces of two different people; and two types of Hemifaces, in which a person's half-face to the right or left was joined to a low-contrast, standard half-face in the opposite visual field. The faces were presented in randomized order for 45 ms followed by a pattern mask. A white stripe along the vertical midline of all faces concealed the discrepancy between the two halves, such that subjects were unaware that some faces were not bilaterally symmetric. Despite a higher level of performance for Hemi-left than Hemi-right faces and higher frequency of left than right hemispatial matches for Chimeric faces, ERPs revealed no difference between Hemi-left and Hemi-right faces or between left and right hemispatial matches of Chimeric faces. The N170 component was larger over the right than the left temporal cortex for all types of faces. Interestingly, ERPs to Bisymmetric faces diverged from those to Hemifaces at 200 ms poststimulus and from those to Chimeric faces at 400 ms. These findings reveal that basic facial properties that are absent in Hemifaces are discriminated earlier in processing than information about facial identity.

INVESTIGATING NEURAL CORRELATES OF ENCODING AND DIRECTED FORGETTING USING EVENT-RELATED FMRI.

Robert M. Siwiec*, Paul J. Reber*, Ken A. Paller*, Darren R. Gitelman+,#, Todd B. Parrish#, and M.-Marsel Mesulam+, Departments of Psychology*, Neurology+, and Radiology#, Northwestern University, Evanston, IL.

Recent functional neuroimaging studies have shown the importance of the medial temporal lobe (MTL) in conjunction with the inferior frontal gyrus (IFG) in memory encoding. The magnitude of activation evoked in IFG and MTL predicted how well study items were later remembered. Differences in activation that predict subsequent memory, Dm effects, have also been demonstrated with event-related potentials using directed forgetting. By instructing participants to either remember or forget, rather than relying on endogenous variability in encoding, we control for distinctiveness differences among stimuli that could have previously been correlated with successful encoding. Particularly distinctive stimuli may evoke additional visual processing that is correlated with subsequent memory. Participants performed five runs each containing 60 study items. Half the study items were cued as "remember" trials and half were cued as "forget" trials. In addition, half the study stimuli were words and half were faces in order to examine stimulus-type effects in memory encoding. Each trial lasted 3 seconds (1500 ms word or face stimulus, 1500 ms remember or forget instruction). The intertrial-interval was varied across trials by including blank trials (33% trial periods were fixation points). Consistent with previous findings, encoding contrasted with directed forgetting elicited greater activity in the right and left anterior hippocampus, left IFG, and parahippocampal gyrus.

QUANTITATIVE ESTIMATION OF REPETITION PRIMING EFFECTS IN FMRI.

Paul J. Reber*, Ken A. Paller*, Darren R. Gitelman+,#, Todd B. Parrish#, & M-Marsel Mesulam+, Departments of Psychology*, Neurology+ & Radiology#, Northwestern University, Evanston, IL.

The second presentation of a visual stimulus produces less activity in ventral occipito-temporal cortex (BA 19/37) compared with initial viewing. Examining the effect of multiple presentations with fMRI provides additional insight into this priming phenomenon. Ten participants observed pictures of objects during fMRI of occipital and temporal regions (1.5T Siemens Vision, 16-24x4mm axial slices, TR=2.0s). Each of 4-5 scanning runs included 30 trials (ISI=12s) containing 4 targets and 6 run-unique non-target stimuli presented 1-8 times in a simple target-detection task (respond via button press: 'yes' to the target object, 'no' otherwise). The stimulus-evoked response was estimated for all non-target trials using individually assessed hemodynamic response profiles (separate run, 5x4mm slices, TR=0.75s). Increased activity was observed throughout occipital and ventral temporal visual areas. On stimulus-repetition trials, reductions in activity were evident across ventral occipito-temporal cortical areas. The relationship between the evoked response and repetitions was fit for linear and exponential functions. The exponential model fit better in many areas, but additional areas exhibited a linear decline in responses over 8 stimulus repetitions, possibly reflecting different effects of priming within components of the visual system. While interpretation of this quantitative technique depends on the linear relationship between neural activity and fMRI, the specific forms of the repetition/response curve provide strong constraints for the development of neurally plausible computational models of priming.

NEURAL CORRELATES OF INTENTIONAL AND INCIDENTAL RETRIEVAL FOR WORDS AND FACES

Stephan G. Boehm and Werner Sommer, Humboldt-Universitiy, Berlin, Germany

During memory retrieval event-related potentials (ERPs) show reliable positive differences between old (repeated) and new stimuli under both intentional and incidental retrieval instructions. Here, the electrophysiological correlates of both types of memory retrieval were compared within a single task for short-term repetitions of common words and famous faces. Intentional and incidental retrieval were operationally defined as repetitions of targets and nontargets in a modified Sternberg task. Replicating previous work, two temporally and topographically distinct memory potentials (revealed as difference potential between the ERP of old and new stimuli) - a parieto-central positivity between 300 and 600 ms and an earlier frontopolar positivity between 200 and 400 ms - were observed. The early frontopolar positivity was indistinguishable for both kinds of retrieval but as compared to similar previous experiments where the non-targets had been shown before, its amplitude was reduced. This was caused by some ERP positivity present already at first stimulus presentation, possibly related to encoding processess in the service of subsequent nontarget recognition. In contrast, the late parietal positivity was smaller for incidental than for intentional retrieval for both words and faces. In addition, ERP difference-waves for intentional and incidental retrieval differed in scalp topography indicating contributions of different brain systems. These results indicate that during the first few hundred milliseconds after cue presentation, intentional and incidental retrieval relate to similar neural activity; only thereafter do they start to differ.