Teton Village, Jackson Hole, Wyoming
                             January 30 - February 4, 2005
        Organizer:  George Sperling, University of California, Irvine



Merav Ahissar
Hebrew University, Jerusalem

Low Level Attributes and the Clarity of Perception
Authors:  Merav Ahissar, Ariel Rokem, Israel Nelken and Mor Nahum

  To what extent is a small increase in stimulus level important for clear
perception?  Intuitively, we reason that, under daily conditions, when stimuli
we attempt to perceive are embedded in context, perceptually distinct, and
familiar, perception will not be sensitive to stimulus contrast.  Contrast
elevation will have a greater impact on discriminating between perceptually 
similar elements. This reasoning predicts that the magnitude of the contribution
of binaural interactions to speech intelligibility (measured by the Binaural 
Intelligibility Level Difference, BILD), will be greater when discriminating 
between similar stimuli (e.g. /barul/ vs. /parul/) compared with very distinct 
words (e.g. /dilen vs. /talug/).  We assessed this prediction by applying an 
adaptive procedure to measure thresholds for 80% correct identification when 
noise was in-phase in the two ears, and stimulus was either in-phase or 
anti-phase.  BILD is the difference between these thresholds.
  Contrary to our prediction, we found a large effect in the opposite direction.
Namely, BILD was substantially higher when identification between very different
words was required (~8dB) compared with discrimination between similar words 
(~3dB).  We propose that the former is the ecologically relevant case, since in
a conversational context, we need to decipher between semantically, but 
typically not perceptually, similar words.  Thus, in noisy environments when
perception in not way above threshold, a small addition to low-level cues 
results in substantially increased perception.


Benjamin T. Backus
University of Pennsylvania

Recruitment of New Perceptual Cues
Authors:  Haijiang Qi, Ben Backus, et al.

  To optimally exploit the statistical structure in signals produced by the 
visual environment, the visual system ought to learn when a signal that was 
previously uninformative becomes informative for inferring a property of the
world.  If the visual system can learn to use a newly reliable signal, one would
expect to see effects of the new signal on visual percepts.  We tested this
proposition in the laboratory.  Observers judged the direction of rotation
(about a vertical axis) of a stereoscopically defined cube during a one-hour 
training procedure.  The direction of rotation was made contingent on one of 
three new "cues": the position of the cube within the visual field (above or 
below fixation), the direction of the cube's movement (upwards or downwards), or
a sound (high or low tone).  Ten percent of trials during training were probe 
trials, on which the stimulus was presented monocularly.  After training, the 
two visual cues strongly biased the perceived direction of rotation, but the 
auditory cue did not.  For some observers, learning was gradual over the course
of the training hour.  Learning persisted into the next day.  A second session,
in which the contingency was reversed, was effective at reducing the learned 
bias, but learning of the reversed bias was not as strong as the original.  
These results suggest that visual perception is kept accurate by processes that
actively seek to learn the statistical contingencies that relate measured 
signals to appropriate representations of the world.


Geoff Boynton
Salk Institute

Global Feature-Based Attention for Motion

  Recent physiological and functional neuroimaging experiments show that 
attention to a particular direction of motion can enhance the response to
unattended stimuli sharing the same direction of motion.  Here, we present 
psychophysical evidence of this 'global-feature-based' attentional mechanism 
by assessing the response to an unattended stimulus with the motion aftereffect
  Trials consisted of an adapting phase and a test phase.  During the adapting 
phase, subjects were instructed to perform a two-interval-forced-choice speed 
discrimination task on one of two overlapping fields of upward and downward 
moving dots on one side of fixation.  A single field of upward, downward, or 
uncorrelated field of dots was presented on the unattended side of fixation.  
During the test phase, the MAE induced by the previously unattended stimulus was
measured by having subjects determine the direction of motion of a field of
slowly moving dots placed at the previously unattended location. 
  We found that the MAE induced by the unattended stimulus during the test phase
was much stronger when it moved in the same direction as the attended field of 
dots.  Also, an unattended stimulus that had uncorrelated motion induced an MAE
in the direction opposite to the attended direction of motion.  These results 
show that the response to an unattended moving stimulus can be affected by the 
direction of motion attended to elsewhere in the visual field. 


Elizabeth Brannon
Duke University

The Evolution and Development of Numerical Abilities

  What are the evolutionary and developmental precursors of human mathematical 
ability?  In this talk I will review my research program which aims to map out
the similarities and differences between animal, human infant and adult human 
number representations.  I will draw upon a variety of behavioral and 
neurobiological methods that I have used to probe the numerical representations
of rhesus monkeys, prosimian primates, human infants, human children, and adult
humans. The main argument championed will be that there is a nonverbal mechanism
for representing number as analog magnitudes that has both phylogenetic and 
developmental precursors.


Scott Brown
University of California, Irvine

A Simplified Complete Model of Absolute Identification
Authors:  Scott Brown, Tony Marley, Andrew Heathcote, and Yves Lacouture

  Absolute identification (AI) involves associating labels with stimuli that 
vary on some attribute.  There are many robust empirical benchmarks in AI, eg:
accuracy varies with set size and stimulus size; response time distributions 
have characteristic shapes, and these vary with stimulus size; there are 
characteristic 'bow patterns' in accuracy and sensitivity to different stimuli;
there are also large sequential effects.  No current model of AI accommodates 
all these benchmark phenomena.  We present a model that combines elements of 
three previously successful but incomplete models:  Marley & Cook's 'rehearsal'
model of stimulus representation; Lacouture & Marley's 'bow mapping' model of 
identification; and Brown & Heathcote's 'ballistic accumulation' model of 
response selection.  The model is kept very simple by using only the most 
tightly constrained elements of each model part.  This model is shown to 
accommodate all benchmark phenomena.  It is also simpler than extant models.


Tom Busey
Indiana University

The Role of Configural Processing in the Development of Visual Expertise

  Fingerprint examiners spend long hours exposing their visual systems to a very
specific set of visual stimuli.  Novices have almost no experience with 
fingerprints at the individuating level.  Thus fingerprint experts represent an
almost ideal case in which to address the effects of perceptual learning and 
visual expertise.  In two experiments we examine how experts differ from 
novices.  In an X-AB matching task with fingerprint fragments, experts 
demonstrated better overall performance, immunity to longer delays, and evidence
of configural processing when fragments were presented in noise.  Novices were 
affected by longer delays and showed no evidence of configural processing.  In a
second experiment, upright and inverted faces and fingerprints were shown to 
experts and novices to look for electrophysiological evidence for expertise. 
The N170 EEG component was reliably delayed over the right parietal/temporal 
regions when faces were inverted, replicating an effect that in the literature 
has been interpreted as a signature of configural processing.  The inverted 
fingerprints showed a similar delay of the N170 over the right parietal/temporal
region, but only in experts, providing converging evidence for configural 
processing when experts view fingerprints.  Together the results of both 
experiments point to the role of configural processing in the development of
visual expertise, possibly supported by idiosyncratic relational information 
among fingerprint features.


Barbara Dosher
University of California, Irvine

Perceptual Learning


James Elder
York University

Testing Linear & Nonlinear Detection Models Using Classification Image Analysis
Authors:  James Elder and Yaniv Morgenstern

  Detection of low-contrast luminance-defined stimuli can involve spatial 
summation over a large portion of the visual field.  Prior psychophysical
results, however, suggest that the summation region may shrink substantially in 
the presence of high-contrast masking gratings or noise (Legge & Foley, 1980; 
Kersten, 1984).  This may be related to recent findings that the receptive 
fields of V1 neurons contract as stimulus contrast increases (Sceniak et al., 
1999).  Here we use a classification image technique to directly test whether 
the psychophysical receptive field for a simple stimulus (a vertical edge in 
noise) is dependent upon contrast.  Classification images for edge detection 
were estimated at noise contrasts ranging from 4%-50%.  Estimated receptive 
fields were found to be well-approximated by elongated 2D Gaussian derivative or
Gabor filters.  Filter width was found to depend strongly on contrast, but not 
as predicted:  summation fields were found to expand by a factor of 2-3 as 
contrast increased.
  A second surprising result is that the estimated summation fields for edge
detection are very long: up to about 25 deg (full length at half height).  This
means that detection must be based upon spatial pooling over many neurons in 
striate cortex, given that V1 simple cells are typically on the order of 1 deg 
in length.  While linear pooling is normally assumed in classification image 
estimation, nonlinear pooling (e.g. probability summation) could yield an 
identical classification image. Here we use a doublepass technique to explicitly
test both linear and nonlinear pooling models. Models are constrained to account
for 1) the estimated classification image, 2) human performance, and 3) human 
consistency, and then evaluated in terms of trial-by-trial correlation with 
human response data.  Consistency between the linear model and the human data 
was better than for nonlinear models, and almost as good as human within-subject
consistency, indicating that the linear model accounts for virtually all 
deterministic aspects of detection.


Vincent Ferrera
Columbia University

Functional Imaging of Categorical Decision Processes
Authors:  V. P. Ferrera, J. G. Grinband, and J. Hirsch

  Visual information processing involves making decisions at many different 
levels of abstraction, from detection to discrimination and categorization.  At
the more abstract levels, decisions require integration of sensory data with
context-specific information.  Context can be used to establish "internal
standards," "categorical boundaries," or more generally, "decision criteria" for
classifying stimuli.  To study this issue we have developed a novel 
psychophysical categorization task in which the subject's decision criterion
varies while the stimulus set, judgment type and motor responses remain fixed.
In this task, subjects categorize a set of stimuli that vary continuously along
a single dimension (length of lines or speed of moving dots).  On each trial, a
cue instructed the subjects where to set their decision criterion for 
categorizing the stimulus as "short/long" or slow/fast."  Depending on the cue,
subjects change the interpretation of a stimulus while making the same type of 
judgment.  Behavioral analyses indicate that this task minimizes the effects of
previous trials, and allows one to dissociate neural activity related to setting
the decision criterion from activity related to response selection.  Subjects
performed the task during event-related fMRI scanning sessions to determine
which areas are active during the generation or shifting of the decision 
criterion.  There were two control tasks.  The first was a two-alternative 
discrimination task to control for perceptual processing unrelated to criterion
generation.  The second was a delayed match-to-sample task to control for
activity related to visual working memory.  All tasks were equated for 
difficutly and performance level.  We have identified decision-related 
activation in the medial prefrontal gyrus, dorsolateral prefrontal, and inferior
frontal cortices.  These results suggest areas that may be involved in decision
making in addition to those that have been identified by neurophysiologcal 


Jackie Gottlieb
Columbia University

Integration of Attention and Motor Planning in Monkey Posterior Parietal Cortex

  In classical neuroscience, vision and attention are generally considered 
separately from movement planning.  Insofar as it has been investigated, the 
relation between attention and motor planning has only been considered in the 
forward direction: attention selects the target of a directed movement, such as
an eye movement or a reaching movement, and this selection is a precursor to
movement planning.  The possibility of more complex, or even backward 
interactions between movement planning and attention is much more rarely 
considered.  However, in the present experiments we show evidence of just such 
interactions in monkey posterior parietal cortex.  We examined neural responses
in a portion of the monkey posterior parietal cortex - the lateral intraparietal
area (LIP) that had been considered important for attentional orienting and/or
planning for rapid eye movements (saccades).  The monkey performed a visual
search-and-discrimination task in which he had to find a predefined target in 
its visual periphery, without using gaze shifts, and to indicate his 
discrimination of target shape in a 2-alternative forced choice, reaction-time 
procedure, by briefly releasing grasp of a bar with the right or left hand.
Consistent with the idea that LIP neurons encode the locus of attention, nearly
all neurons showed spatial tuning for the location of the target.  Unexpectedly,
however, the vast majority of these neurons were also strongly affected by the 
manual response.  Neurons responded strongest if the search target was in their
receptive field (RF) AND the monkey released the hand ipsilateral to the 
receptive field.  In general, responses for congruent configurations (e.g., 
target on the right and right-hand release or target on the left and left-hand 
release) were stronger than those for incongruent configurations (target on the
right and left-hand release or vice-versa).  When compared across two monkeys 
the relative magnitude of the neural congruence effects mirrored the relative 
magnitude of the behavioral congruence effects.  The results show that LIP 
neurons carry attentional signals that are not obligately linked to saccade 
preparation but are nevertheless modified by motor planning.  At the level of 
parietal cortex, therefore, attention may be inextricably linked to motor 
planning.  Moreover, the results show that posterior parietal cortex is not as
tightly compartmentalized on the basis of effector specificity as is commonly

Emily Grossman
University of California, Irvine

STSp and Biological Motion Perception: An rTMS Study
Authors: Emily Grossman, Lorella Battelli & Alvaro Pascual-Leone

  Neuroimaging studies have identified cortical activity on the posterior 
extent of the superior temporal sulcus (pSTS) to be correlated with the 
perception body movements, including face, hand and body movements.  Robust 
neural activity can be observed within pSTS even during observation of
point-light biological motion, animations in which actions are portrayed by a 
dozen light points located on the joints.  In this study we evaluated whether an
intact pSTS is necessary for the perception of biological motion in individuals
with otherwise normal brain function.  We have used low frequency repetitive 
(1 Hz for 10 min) transcranial magnetic stimulation (rTMS) to temporarily 
disrupt cortical functioning within a localized neural region on the pSTS.  
Observers discriminated between point-light biological and motion-matched 
nonbiological controls ("scrambled" motion) embedded in dynamic noise arrays. 
To test for a possible generalized effect of brain stimulation, observers also 
discriminated upside-down biological animations from upside-down scrambled 
motion.  Biological motion sensitivity (upright and inverted) was measured under
normal viewing conditions, immediately following brain stimulation over the 
right pSTS, and immediately following stimulation over left MT+/V5 (a control 
  We found biological motion sensitivity to be temporarily impaired following 
low frequency stimulation over pSTS, and fully recovered following a 15 minute 
rest period.  Inverted biological motion sensitivity was unaffected by 
stimulation over STSp.  We were surprised to find that stimulation over MT+/V5
did not affect observer sensitivity in either task (upright or inverted), 
despite neuroimaging studies clearly demonstrating MT+/V5 to be activated - 
albeit nonselectively - by biological motion.  These results confirm the 
importance of the posterior superior temporal sulcus in biological motion 


David J. Heeger
New York University

Functional Organization of Human Posterior Parietal Cortex

  It is widely believed that neurons in certain parietal and prefrontal cortical
areas control visual attention.  A number of these cortical areas have been 
defined with precision in the monkey brain, but less so in the human brain. 
Decades of research in visual cortex has taught us that there is little hope of
understanding the function of these cortical areas without first having precise,
routine, and reliable methods for defining them.  This is self-evident from an 
experimental point of view; one must be confident about performing the same  
measurement repeatedly in the same bit of cortical tissue for the necessary
replications and controls.  Cortical areas are defined using a variety of
methods and a confluence of factors.  One of the primary factors is topography,
which refers to the orderly layout of information across a cortical area.  For
example, it has been known for nearly 100 years that there are multiple 
representations of the visual field in visual cortex, each of which corresponds
to a separate and distinct visual cortical area.  fMRI has proven to be the best
current method for measuring and characterizing topographic maps.  Topography is
not, however, restricted to primary sensory areas and may be an organizing 
principle common to most, if not all, of the cerebral cortex.
  Functional magnetic resonance imaging (fMRI) was used to measure activity in 
human parietal cortex during performance of a visual detection task in which the
focus of attention systematically traversed the visual field.  Critically, the
stimuli were identical on all trials (except for slight contrast changes in a 
fully randomized selection of the target locations) while only the cued location
varied.  Traveling waves of activity were observed in posterior parietal cortex
consistent with shifts in covert attention in the absence of eye movements.  The
temporal phase of the fMRI signal in each voxel indicated the corresponding 
visual field location.  Visualization of the distribution of temporal phases
on a flattened representation of parietal cortex revealed two distinct 
topographically-organized cortical areas within the intraparietal sulcus (IPS1 
and IPS2), each representing the contralateral visual field. IPS1 shares a lower
vertical meridian boundary with the dorsal extent of cortical area V7, and its
upper vertical meridian representation is shared with the ventral boundary of
IPS2. Both IPS1 and IPS2 exhibited little response to passive visual stimulation
relative to early visual areas.  These results provide the first evidence for
topographic organization of attention-related signals in human parietal cortex.


Holly Jimison
Oregon Health and Science University

Unobtrusive Monitoring of Computer Interactions to Detect Cognitive Change
Authors:  Holly Jimison, Misha Pavel, & James McKanna

  The United States has experienced a rapid growth in the use of computers by 
elders.  Email, Web browsing, and computer games are among the most common 
routine activities for this group of users.  Researchers have demonstrated the 
importance of the early detection of cognitive decline.  Users over the age of 
75 are at risk for medically related cognitive problems and confusion, and early
detection allows for more effective clinical intervention.  As part of an effort
to develop new techniques for the early detection of cognitive decline in 
elders, we have developed unobtrusive monitoring techniques using computer/user
interactions to detect sustained cognitive change.  We will describe our 
algorithms for inferring clinically significant trends in a user's cognitive 
performance using monitoring data from computer games and psychomotor 
measurements associated with keyboard entry and mouse movement.  In addition to
using these measures to classify significant performance changes, we also adapt
computer interfaces with tailored hints and assistance when needed.  We will 
describe our pilot tests of this research approach in a group of elders in a 
senior residential facility.


Christopher T. Kello
George Mason University

1/f Noise in the Intrinsic Fluctuations of Human Behavior
Authors:  Christopher T. Kello and Brandon C. Beltz

1/f noise in fluctuations of human behavior has been attributed to a variety of
processes, but these accounts have not addressed its ubiquity in biological 
systems, and complex systems more generally.  Four experiments are reported in 
which 1/f noise in series of key-press responses is selectively whitened as a 
function of cue predictability.  In the first three experiments, 1/f noise in 
response latencies was more whitened in blocks of unpredictable cues, compared 
with blocks of predictable cues.  By contrast, 1/f noise in key-press durations
was equally evident in these blocks.  In a fourth experiment, predictable and
unpredictable cues were randomly interspersed in series, yet the same difference
in whitening was found. We argue that the observed patterns of whitening lead to
unexpected and seemingly implausible sets of long-range memories or processes.
We propose instead that 1/f noise permeates background variability of human 
behavior, and that this background is observed most clearly when behavior is 
measured repeatedly and consistently, with minimal perturbation and minimal 


Peter Lennie
New York University

The Color Mechanisms in Striate Cortex
Authors:  Peter Lennie, Chris Tailby, Sam Solomon, Neel Dhruv, and Najib Majaj

  Psychophysical work points very clearly to the special status of three 
post-receptoral mechanisms of color vision, two color-opponent ones tuned 
respectively to red-green and yellow-blue axes of color variation, and a third 
non-opponent one tuned to light-dark (achromatic) variations.  Physiological
work in early visual cortex has so far not revealed counterpart chromatic 
mechanisms: the chromatic preferences of neurons do not cluster around the 
'cardinal' axes of color space.  Using adaptation to chromatic contrast we have
been able to expose in V1 of macaque a pair of chromatic mechanisms whose 
properties are well-aligned with those of the cardinal mechanisms identified 
psychophysically.  These cardinal mechanisms probably arise at the earliest 
stage of cortical input, and signals from them drive the chromatic response of 
a V1 neuron, though their separate contributions to the neuron's chromatic 
tuning are not usually discernable.


Geoff Loftus
University of Washington

Seeing Things at a Distance

  It is a matter of common sense that an object is easier to recognize when
close than when far away.  A possible explanation for why this happens begins
with two observations.  First, the human visual system, like many
image-processing devices, can be viewed as a spatial filter which passes higher
spatial frequencies, expressed in terms of cycles/degree, progressively more 
poorly.  Second, as an object is moved further from the observer, the object's 
image spatial frequency spectrum expressed in terms of cycles/object scales 
downward in a manner inversely proportional to distance.  An implication of 
these two observations is that as an object moves away, progressively lower 
spatial frequencies, expressed in cycles/object=8Band therefore progressively 
coarser object details=8Bare lost to the observer at a rate that is likewise 
inversely proportional to distance.
  We propose what we call the distance-as-filtering hypothesis, which is that
these two observations are sufficient to explain the effect of distance on
object processing.  If the distance-as-filtering hypothesis is correct then one
should be able to simulate the effect of seeing an object at some distance, D, 
by filtering the object so as to mimic its spatial-frequency composition, 
expressed in terms of cycles/object, at that distance.  In four experiments we 
measured perception of two classes of object--faces or vehicles--at varying 
distances that were simulated either by filtering the object as just described,
or by shrinking the object so that it subtended the visual angle corresponding 
to the desired distance.  The distance-as-filtering hypothesis was confirmed 
perfectly in two tasks: assessing the informational content of the object and 
object classification.  Data from the two tasks could be accounted for by 
assuming that they are mediated by different low-pass spatial filters within the
human visual system that have the same general mathematical description, but 
that differ in scale by a factor of approximately 0.75.


Zhong-Lin Lu
University of Southern California

Deficits in Forming Perceptual Templates May Underlie the Etiology of 
  Developmental Dyslexia 
Authors: Zhong-Lin Lu, Anne J. Sperling, Franklin R. Manis, & Mark S. Seidenberg

  Formation of optimal phonological and orthographical templates is critical for
speech perception and reading.  A general deficit in forming perceptual 
templates distorts speech perception in infancy, retarding development of 
phonological categories.  It can also affect letter recognition and encoding of
letter patterns and sequential redundancies.  The behavioral signature for 
non-optimal perceptual templates is reduced ability in processing information 
embedded in high external noise ("TV snow"), compared to "normal" behavior in 
processing "clean" signals (Lu & Dosher, 1998).  In this study, we compared 
contrast sensitivity of dyslexic and non-dyslexic children using sine wave 
gratings designed to activate either magnocellular (M) or parvocellular (P) 
processing.  The gratings were either displayed without noise, or embedded in a
noise patch.  Dyslexics had higher contrast thresholds than non-dyslexics when 
the gratings were displayed in high noise, in both the M and P versions.
Dyslexics performed as well as non-dyslexics, however, when the gratings were 
displayed without noise, again in both M and P versions.  In addition, contrast
thresholds in high external noise conditions correlated with language measures,
as well as word reading and orthographic measures.  Dyslexics with language 
impairments tended to have the highest thresholds.  The same pattern of results
was obtained in several related studies using different tasks.  Our results 
suggest that dyslexic children may have a general deficit in forming perceptual
templates instead of having just magnocellular deficits.  The inability to form
optimal perceptual templates primarily impairs language development, which may 
in itself contribute to deficits in phonological processing. 


Kenneth Malmberg
Iowa State University

A Signal Detection Analysis of Ratings and Remember-Know Judgments


Jeffrey B. Mulligan
NASA Ames Research Center

Taking Gaze Tracking from the Lab to the Field (and Sky)

  Gaze tracking measures can provide information about the spatial locus of
attention of a behaving agent.  Here we examine the looking behavior of 
helicopter pilots flying under visual meteorological conditions.  The goal of 
the study is to correlate various types of looking behavior with pilots' 
accuracy in maintaining a precisely specified route, to support the formulation
of new regulations and procedures.  Eight pilots were instructed to fly a 
precision route specified by a series of waypoints.  The geographic coordinates
of the waypoints were entered into an onboard receiver of global positioning
system (GPS) signals.  Using a single 8mm videocassette, we recorded four video
streams (30 frames per second), one audio stream, and GPS data sampled at 1 Hz.
The four video streams were comprised of two cameras attached to the pilot's 
head, and two stationary cameras mounted on the aircraft.  The head-mounted
cameras consisted of a camera viewing the pilot's right eye through an infrared
"hot" mirror, and a forward-looking scene camera located in front of the 
subject's forehead.  From the eye images we compute estimates of head-relative 
gaze, while we obtain independent estimates of the head pose from the 
head-mounted scene camera and a stationary "face" camera.  This talk will 
present an overview of the technical challenges encountered in the processing of
the images, as well as preliminary results of the study.


Clark Ohnesorge
Gustavus Adolphus College

About Face: Hemispheric Effects in the Recognition of Self
Authors:  Ohnesorge, Johnson, and Palmer

  We present data from several studies investigating hemispheric specialization
for self-recognition.  Researchers have recently addressed this issue using 
techniques ranging from behavioral responding to neuroimaging during 
self-recognition, often using stimuli created by morphing together each 
subject's face with that of a familiar other.  Generally split-visualfield
presentation is used to support inferences about lateralization.  While previous
studies have shown changes in response probability (i.e. the classification of a
particular stimulus as either "self" or "other") as a function either of visual
field or response hand, it has remained unclear whether those changes are due to
differences in perceptual sensitivity or decisional bias.  In our first study we
presented individualized morphed stimuli using a split-visualfield technigue and
using a detection design (theory of Signal Detectability) to reveal large 
effects in both Sensitivity and in Bias.  In each case the effect was an 
interaction between Visual Field and Response Hand.  In a second study we used a
2afc task to eliminate bias and yield a cleaner, but somewhat different view, of
the Sensitivity effect we observed in our first study.


Carl Olson
Carnegie Mellon University

Reward-Related Activity in Frontal Cortex:  Does it Represent Value or Reflect 

  In numerous brain areas, neuronal activity varies according to the size of 
the reward for which a monkey is working.  Reward-dependent activity has
commonly been viewed as representing the value of the reward.  Alternatively, 
however, it could reflect the monkey's degree of motivation.  Anticipation of a
more valued reward leads to stonger motivation as evidenced by measures of 
arousal, attention and intensity of motor output.  We have distinguished between
value-related and motivation-related processes in single-neuron recording
studies of monkeys working to obtain rewards and avoid penalties.  We have found
that the nature of reward-dependent activity varies across areas in the frontal
lobe.  Neuronal activity in orbitofrontal cortex genuinely represents the value
of the anticipated outcome.  In contrast, neuronal activity in other frontal 
areas is determined by the monkey's degree of motivation.  Our findings cast 
light on the stages by which representations of goal value (in the limbic
system) are transformed into the motivated pursuit of goals (in sensorimotor 


Tatiana Pasternak
University of Rochester

Directional Signals in Prefrontal Cortex During a Working Memory for Motion Task

  During the performance of working memory tasks neurons in prefrontal cortex 
(PFC) respond to sensory stimulation and show stimulus selective activity during
the memory delay.  Prefrontal cortex is interconnected with a number of areas in
the dorsal visual stream, including area MT associated with processing and 
remembering of visual motion.  We investigated the behavior of PFC neurons 
during a task in which the monkeys compared the direction of motion in two 
random-dot stimuli, sample and test, separated by a brief memory delay.
Responses of nearly 80% of recorded neurons showed some degree of direction 
selectivity and were affected by the coherence level of the motion stimuli.  The
nature and the temporal dynamic of these responses to motion suggest that this
information may arrive from area MT and/or other cortical neurons processing 
visual motion.  Throughout the memory delay over 20% of recorded neurons carried
reliable information about the remembered direction of motion and this number 
increased to 40% at the end of the delay, immediately prior to test onset.  For
the majority of neurons the directional signal during the delay was transient 
(< 500ms) and distributed among many neurons.  This delay activity was 
behaviorally relevant since on error trials the directional signal decreased 
towards the end of the memory delay and was largely absent at the time of test 
onset.  Our results provide evidence that PFC neurons participate in the
processing and short-term storage of information used in the working memory for
motion task.  Furthermore, these data support a dynamic model of working memory,
in which stimulus selectivity is preserved reliably across a population despite
the fluctuation of firing rates of individual neurons throughout a memory delay.


Misha Pavel
Oregon Health and Science University

Model-Based Unobtrusive Monitoring of Elders' Activities and Mobility to Assess
  Cognitive Function
Authors:   Misha Pavel, Holly Jimison, Tamara Hayes & Jeff Kaye

  This talk describes a model-based approach to the interpretation of 
unobtrusive sensor data used to monitor elders in their home environment to 
assess their health, daily activities, and cognitive function.  We will present
a computational approach based on a semi-Markov model that provides a link 
between the sensor data and the participants' movements.  The parameters of the
model that characterize an elder's activities and mobility in the home 
environment are learned using an unsupervised procedure.  We will discuss our 
plan to examine the hypothesis that the resulting model parameters can predict 
future cognitive functionality and health state of the elders.  The ultimate 
goal is to develop intelligent support systems that would enable elders to live
independently in their home environment.


Alex (Sandy) Pentland

Social Signals

  Nonlinguistic social signals (e.g., 'tone of voice') are often as important as
linguistic content in predicting behavioural outcomes [1,2].  This paper 
describes four automated measures of such social signalling, and shows that they
can be used to form powerful predictors of objective and subjective outcomes in
several important situations.  Finally, it is argued that such signals are 
important determinants of social position.


Michael Platt
Duke University

Representation of Subjective Utility in the Primate Brain

   Decision-making models derived from economics, psychology, and behavioral
ecology propose that organisms choose based on value.  Recent neurological 
studies suggest that neurons in several brain areas linking visual perception
with orienting track fluid rewards associated with visual targets.  Here we show
that neurons in cingulate and parietal cortex track subjective orienting biases
for targets associated with uncertain rewards and the opportunity to view 
preferred social images, respectively, even when target value is held constant.
These data suggest that orienting decisions are made by scaling neuronal 
responses to targets by their subjective utility.

John Reynolds
Salk Institute

Surface-Based Attentional Selection Across Multiple Levels of Visual Processing
Authors: Jude F. Mitchell, Gene R. Stoner, John H. Reynolds

  Binocular rivalry and attention both involve selection of visual stimuli, but
affect perception quite differently.  During binocular rivalry, awareness 
alternates between two different stimuli presented to the two eyes.  In
contrast, attending to one of two different stimuli impairs discrimination of 
the ignored stimulus, but without causing it to disappear from consciousness.
Here we show that, despite this difference, attention and rivalry rely on shared
object-based selection mechanisms.  We cued attention to one of two superimposed
transparent surfaces and then deleted the image of one surface from each eye, 
resulting in rivalry.  Observers usually reported seeing only the cued surface.
They were also less accurate in judging unpredictable changes in the features of
the uncued surface.  Our design ensured that selection of the cued surface
could not initially have been triggered by spatial, ocular or feature-based
mechanisms.  Rather, attention was drawn to one surface, and this caused the 
other surface to be perceptually suppressed during rivalry.  However, following
the initial surface-based selection, the impairment in judging the uncued 
surface was strongly influenced by interocular competition, demonstrating 
interacting selection across levels of visual processing.


Cassandra Schroeder
Arizona State University East

Effects of Incentive on Action Representation: A Study of Gender Dyads 
Performance in Simple Compatibility Tasks

Authors:  Cassandra Schroeder and Rob Gray

  It has been previously shown that individuals working in dyads represent the 
actions performed by the other person when performing a simple compatibility 
task even when they interfere with their own task (Sebanz, Knoblich & Prinz, 
2003).  The purpose of this study was to examine how the dyad gender and task 
incentive effect this action representation.  For this study all possible 
combinations of gender in the dyads were observed.  The two incentive conditions
involved individuals in the dyad receiving a reward based upon their individual
performance or based upon collaborative performance.  Speed and accuracy 
incentive criteria were constructed based upon a previous study that was used as
a baseline.  Previous research has shown at least two different relationships 
between incentive and performance. One line of research proposes that incentives
may backfire and reduce performance (Fehr & Falk, 2002), while the other line of
research states that incentive may stimulate and higher spontaneous goal setting
and increase the performance in a group (Guthrie & Hollensbe, 2004).  Other 
research has shown that men and women have distinct communication styles.  Men
seem to be more direct and debate-like, while women show more positive 
politeness and consideration (Savicki, Kelley and Ammon, 2002).  Our first 
prediction was that individuals working in dyads would represent otherıs actions
to a greater extent (as evidenced by larger compatibility effects) when the 
incentive is collaborative because there will be more motivation to perform and
represent otherıs actions.  We also predicted that same gender dyads would show
a larger compatibility effect caused by the similarity in communication styles 
shared by a same gender dyad.  Therefore, the compatibility effect is expected 
to be the most pronounced in the collaborative incentive condition where the 
dyad is consisting of the same gender. The results indicated that these proposed
hypotheses were supported indicating that social factors can influence the 
representation of actions.


Shihab Shamma
University of Maryland

Auditory Streaming and the Spectrotemporal Analysis of Sound in the Cortex
Authors:  Shihab Shamma and Mounya ElHilali

  Auditory streaming is a phenomenon that manifests itself in the everyday 
ability of humans and animals to parse complex acoustic information arising from
multiple sound sources into meaningful auditory "streams."   For instance, the 
ability to follow a conversation at a noisy cocktail party or hear the violin in
the orchestra both rely on the formation of auditory streams.  While seemingly 
effortless, the neural mechanisms underlying auditory streaming remain a
  In this talk, we shall discuss how this perceptual ability may emerge as a 
consequence of the multiscale spectrotemporal analysis of sound in the auditory
cortex.  A simplified model of this process is developed to demonstrate how 
auditory streaming could underlie the separation of complex sounds from two 
sources (e.g., speech or music).  The model reveals that certain perceptual 
attributes are critical for the simultaneous separation of sounds (pitch and 
onset sensiticity), whereas others are important for the sequential binding of 
the segregated sound features (timbre and location).


S. Murray Sherman
University of Chicago

Some Thoughts on the Functioning of the Thalamus

  The thalamus has long had a bad press, seen as a simple, machine-like relay of
information to cortex.  Work on the visual thalamic relays provides two key 
properties that has dramatically changed this view.  First, ~95% of input to LGN
relay cells is nonretinal and modulates the relay in dynamic and important ways
related to behavioral state, including attention.  Much of this is related to
control of a voltage-gated, low threshold Ca2+ conductance that determines 
response properties of relay cells and thus affects the very nature of 
information relayed.  Second, the LGN and pulvinar (a massive but generally 
mysterious and ignored thalamic relay), are examples of two different types of 
relay: the LGN is a first order relay, transmitting information from a 
subcortical source (retina), while the pulvinar is mostly a higher order relay,
transmitting information from layer 5 of one cortical area to another area.
Higher order relays seem especially important to general corticocortical 
communication, and this view challenges the conventional dogma that such 
communication is based on direct corticocortical connections.  In this sense, 
any new information reaching a cortical area, whether from a subcortical source
or another cortical area, benefits from a thalamic relay.  Other examples of 
first and higher order relays also exist.  Thus the thalamus not only provides a
behaviorally relevant, dynamic control over the nature of information relayed,
it also plays a key role in basic corticocortical communication.


Steve Shevell
University of Chicago

Color Shifts Caused by Perceptual Grouping

  The perceived color of a light depends on other light nearby.  This is the
well known phenomenon of chromatic induction, which has been studied for more 
than 170 years.  A much more recent observation (AIC 2004) is that chromatic 
induction in one region of the visual scene carries over to a separate, remote 
region that belongs to the same perceptual group.  In experiments, a test square
was at the center of an "hourglass" structure formed with distant stripes also 
in view.  The test square shifted in color toward the induced appearance of the
distant stripes, whose color was affected by local chromatic induction. 
  The color shift declined with weakened grouping established by altering the 
distant stripes so as to disrupt the hourglass structure, by changing either the
shape (static stimulus) or the congruence of motion (moving stimulus) of the 
elements forming the "hourglass."  Also, weakened grouping with the stripes by 
perceiving them in a separate depth plane than the test (via stereo disparity) 
reduced the color shift.  Finally, a binocularly fused "hourglass" with the 
stripes and the test square seen through opposite eyes did shift the color of 
the test.  These results corroborate color shifts caused by perceptual grouping
and implicate a cortical neural substrate.


Richard Shiffrin
Indiana University

What Masking Tells Us About Visual Word Identification

Authors:  Adam Sanborn, Richard Shiffrin and Ken Malmberg

  Masking was used to assess how observers use different types of information to
identify a briefly presented target (e.g., "mouse").  Observers were asked to 
choose between two alternatives.  The target could be identified on the basis of
its case (e.g., a choice between "mouse" and "MOUSE"), by its word identity
(e.g., choices "mouse" and "house"), or with both types of information (e.g.,
choices "mouse" and "HOUSE").  Observers presented brief low-contrast targets 
without a backward-mask used case information more effectively than word 
identity, but a backward-mask (e.g., @@@@) reversed the effect.  More 
particularly, case judgments can be made at very low target contrasts, but are 
adversely affected by masks whereas word identity judgments require higher 
target contrasts, but are less affected by masks.  When both word and case 
distinguish the alternatives, an observer's performance is usually equal to the 
single best source of information, whether the target is masked or not.  We use
these and some related findings to form a tentative and qualitative model of 
visual word processing.


George Sperling
University of California, Irvine

Cue Combination in Third-Order Motion Perception
Authors:  George Sperling and Chia-huei Tseng

  In the Lu-Sperling theory of three orders of motion systems, the third-order 
motion system computes the motion of figures defined on a figure-ground salience
map.  Two motion stimuli have previously been shown to be perceived exclusively
by the third-order motion system: A translating isoluminant red-green grating
and a depth-grating defined by a dynamic random-dot stereogram (DRDS).  Consider
two gratings translating in the same direction: When they stimulate the same 
motion system, perceived motion will increase or cancel depending on the 
relative phase.  When they stimulate different systems, perceived motion is 
always greater than for either grating alone.
  These predictions were tested with a translating, dynamic, random-dot stereo 
grating in which small amounts of isoluminant red or green (in DKL space) were 
added to either the foreground or background stripes.  In 5/6 observers, we 
found a phase in which same-direction color and stereo motion, each of which is
completely visible alone, in combination partially or completely cancelled each
other.  Conclusion: Movements of isoluminant red-green gratings and of 
random-dot stereo depth gratings are perceived by the same salience 
motion-perception mechanism.
  A three-stage motion model consisting of (1) representation of cues in a 
salience map, (2) an elaborated Reichardt Detector to extract motion from 
salience, and (3) a decision rule accounts for 99% of the variance of the data.


Courtney Stein
Dartmouth College

Naive Theories of Proximity:  Misconceptions of Planetary Motion
Authors:  Courtney Stein and Kevin Dunbar

  Students have many misconceptions about the nature of the universe that are 
difficult to eliminate.  One method used to change these misconceptions is to 
pesent students with short videos pointing out the misconception and explaining
the correct model.  Here we conducted two studies to examine the efffects of a
NASA-designed video that describes in detail why the Earth has different
seasons.  The video was presented to undergraduate students who were also given
a test assessing whether this knowledge transferred to a different domain.  We 
found that this short intervention does help students understand superficial
concepts involved in seasonal changes, but NOT the underlying key concepts.
Results are discussed in terms of the competing cognitive processes involved in
conceptual change, which are often ignored in educational interventions with the
net result that the interventions are ineffective.


Mark Steyvers
University of California, Irvine

A Probabilistic Approach for Associative Memory

  To what extent can the acquisition and processing of natural language be 
explained by simple statistical computations?  I will describe a structured 
probabilistic approach to modeling natural language semantics.  The model takes
as input a corpus of words divided into "documents" -- sequences of words that
represent a single coherent discourse, story, conversation, or merely a list. 
It represents each document as the result of drawing words from a set of topics,
where each topic is a probability distribution over words.  The model can be 
used to discover semantically related categories of words, or to identify the 
gist of a document.  The topics in which a word participates reflect the meaning
of that word, and the gist of a document is expressed by the distribution of 
topics that appear in that document.  A set of topics can be learned 
automatically from a collection of documents, as a computational analog of how 
human learners might discover semantic knowledge through their linguistic 
experience.  I will show how the topics model can explain semantic effects in 
word association and episodic memory tasks such as free recall.  Word
association is modeled as a process of predicting which words might occur in the
context of the cue word.  Free recall is modeled as a reconstructive process 
based on verbatim memory traces as well as activated topics that capture the 
gist of a list of words.


Bosco Tjan
University of Southern California

Form Vision in the Periphery
Authors:  Bosco Tjan and Susana T. L. Chung

  Under normal viewing conditions, letters are less visible in the periphery
than in the fovea.  Letter visibility in the periphery decreases further when a
letter is flanked by other letters (“crowding”).  Classical ideal-observer
analysis attributes changes in performance (accuracy or threshold) to two
factors: a change in the equivalent noise internal to the observer and/or a
change in the optimality of the underlying computation, measured in terms of
sampling efficiency.  We extended the ideal-observer analysis to include a
limitation in spatial resolution as defined by the contrast sensitivity function
(CSF) at the eccentricity of interest.  A linear filter, having the shape of the
subject's CSF, was placed in front of an ideal observer.  This is equivalent to
limiting an ideal observer with correlated Gaussian noise instead of white 
noise.  Three observations were made using this CSF-limited ideal-observer 
model.  First, by measuring contrast thresholds for identifying single 
band-passed filtered letters, we found that for letter identification, spatial 
tuning in the periphery (5 and 10 deg eccentricity) was the same as that 
predicted by the CSF-limited ideal-observer model (Chung, Tjan, Legge, 2002, 
Vis. Res.) and was therefore optimal after discounting the CSF.  Second, spatial
tuning remained nearly optimal when the target letter was flanked on both sides
with filtered letters of the same spatial frequency (peak tuning frequency was 
only 0.2 octave higher than that exhibited by the model (Chung & Tjan, 2002, 
ARVO).  Third, the substantial threshold elevation observed in letter crowding
was due to an increase in the subject's equivalent noise, and not to any 
reduction in sampling efficiency (Tjan et al., 2004, VSS).  In addition, we
found that correlated Gaussian noise patches, having the power spectrum of a 
letter, effectively crowded the target letter, suggesting that the sufficient
condition for crowding in the periphery is that the target and its flankers are
close in spatial position and spatial frequencies (Tjan & He, 2004, VSS).


Chia-heui Tseng
Rutgers University

Visual Search Involves Both Sensitization and Suppression
Authors: Chia-huei Tseng, Thomas Papathomas, Zoltan Vidnyanszky, George Sperling

  Tseng et al (2004) reported a week-long sensitization to color based on
attention in a search task.  When a subject is trained for a few days to search
for an object of a particular color, say red, among red, green, blue, and yellow
objects, sensitivity to red is altered for weeks.  We used ambiguous motion 
displays to measure this sensitization, where isoluminant gratings composed of 
red and green stripes alternate in time with texture gratings.  The perceived 
direction of this motion sequence depends both on the attended color and on the
relative saturation of the composing colors.  That is, attending to an 
unsaturated red stripe causes it to behave - in the motion sequence - like a 
more saturated red stripe.  Thus, sensitization can be measured by the 
additional color saturation required to balance the motion sequence.
  Such long-lasting effects could result from either sensitization of the
attended color, or suppression of unattended colors, or a combination of both 
effects.  Here we tried to unconfound these effects by eliminating one of the 
paired colors of the motion display from the search task.  The other paired 
color in the motion display can then be either a target or distracter in the 
search task.  Thereby, we can separately measure the effect of attending to 
target colors on sensitizing the target colors or suppressing distracter colors.
  The motion task indicated both effects: sensitization of the target colors in
the search task and suppression of the distracter colors.  For example, when red
was the attended color and blue was the distracter color, salience to red 
increased at the expense of green for 4/5 subjects.  When blue was the target 
color and red was the distracter color, desensitization of red in favor of green
occurred in 3/4 observers.  The magnitudes of sensitization and desensitization
were positively correlated with search performance.  We conclude that selective
attention to a color in visual search causes long-term sensitization to the 
attended color and long-term suppression of the unattended color; both can be 
accurately assayed by ambiguous motion.