Twenty-Third Annual Interdisciplinary Conference
Teton Village, Jackson Hole, Wyoming
February 1 -- February 6, 1998
Organizer: George Sperling, University of California, Irvine


Herve Abdi
University of Texas at Dallas
P.O. Box 830688, MS:GR4.1
Richardson, TX 75083-0688

Title:  2D or not 2D, that is the question: What can we learn from computational
models operating on 2D representations of faces?

Brief Summary:  Previous work has suggested that computational models of faces
operating on 2D pixel intensity representations can, to some extent, handle
changes in orientation. By using multiple views of a given face instead of a
single view to represent the face, pixel-based representations combined with
simple computational models are able to recognize known faces from new view
angles.  The analysis of the internal representation abstracted by one model
(i.e., a linear auto-associator) showed a spontaneous dissociation between two
kinds of perceptual information: orientation versus identity. Computational
models can also give some insights into the information that can be transferred


Richard B. Anderson
Department of Psychology
Bowling Green State University
Bowling Green, OH 43403

Title:  Division of Memory Effort: New Territory for the Rational Analysis of

Brief Summary:  Extant data indicate that memory trace retention is predicted by
the trace's future need probability.  Thus, retention appears to be rational,
insofar as it is environmentally adaptive.  However, new data suggest that when
responsiveness to need probability requires division of memory effort, the cost
of such responsiveness can nullify potential gains.  The findings have general
implications for rational analysis, as well as practical implications for
optimizing memory performance.


Jo-Anne Bachorowski
Department of Psychology
301 Wilson Hall
Vanderbilt University
Nashville, TN   37240

Title:  Gender and Talker Identity-Cueing in Speech and in Laughter

Brief Summary:  Acoustic representation of gender & talker identity in speech 
can be described by integrating source-filter theory along with knowledge
regarding both sexual dimorphism and unique variation in speech production-
related anatomy. Broadening this perspective, human laughter is conceptualized
as an evolutionarily conserved vocal communication repertoire that functions
largely as an identity-signaling system. Laugh acoustics are expected to vary
in accordance with particular sender-receiver relationship characteristics, such
as gender and relative social dominance. Empirical support for this overall
perspective comes from examining speech acoustics, laugh acoustics in relevant
social contexts, and identity-cueing in nonhuman primates.


Bettina L. Beard
NASA Ames Research Center
Mail Stop 262-2
Moffett Field, CA 94035-1000

Title:  Characterizing internal template representations using response
correlation images.

Brief Summary:  In a discrimination experiment, observers compare two stimuli 
that differ.  Perceptually there is a continuum of image percepts. To make the
discrimination, the observer must classify this continuum into two discrete
categories. The rule that the observer uses to segregate two distinct categories
would depend on the task and the variable that is being manipulated in the
experiment. Here we describe a technique that provides a window to visualize the
underlying stimulus classification rules.  To determine what aspects of a 
stimulus are used to classify the image into a discrete category, random, white
noise is added to the stimulus.  The noise is of very low contrast so that its
effect on threshold is small but still will influence the decision rule.  The
noise presented on each trial, the observer's response ("YES" or "NO") and the
stimulus type (SIGNAL or BLANK) are recorded.  The noises are then averaged for
each of the four stimulus-response categories.  These averaged noises are then
summed to form a classification image illustrating what aspects of the noise
image contributed to the decision rule.  A perceptual classification image is
the correlation over trials between the local noise contrast and the observer's


Geoffrey Boynton
Department of Psychology
Stanford University
Jordan Hall, Bldg. 420
Stanford, CA 94305

Title: Attentional modulation in human V1 measured with fMRI

Authors: Geoffrey M. Boynton and David J. Heeger
Brief Summary:  The purpose of our study is to test if spatial attention
modulates neural activity in human primary visual cortex.  Subjects performed a
speed discrimination task while fMRI signals were recorded using a T2* weighted
spiral acquisition.  Stimuli were moving gratings restricted to a pair of
peripheral, circular (3 deg diameter) apertures, centered 7 deg to the left and
right of a central fixation cross.  Each trial consisted of two 750 msec
stimulus intervals.  Within each aperture, a baseline speed was presented during
one interval and a slightly faster test speed was presented during the other
interval.  Subjects were cued (by a slight modification to the fixation cross)
toward either the left or right aperture and instructed to pick the interval
with the faster speed.  Although the speed of the uncued stimulus also varied,
it was irrelevant for a correct response because both the baseline speeds in the
two apertures and the order of baseline/test in the two apertures were
independently randomized.  The speed increment was chosen based on separate
psychophysical measurements so that subjects would perform with an accuracy of
approximately 78% correct.  The cue alternated between left and right every 18
seconds.  FMRI response was quantified as the phase and amplitude of the (36
sec period) sinusoid that best fit the average time-series of pixels in V1.  The
data were analyzed separately for each hemisphere.  Area V1 was defined in each
hemisphere of each subject's brain using standard techniques that locate
reversals in the retinotopic map of visual cortex.  V1 responses in each
hemisphere modulated with the alternation of the cue; responses were greater
when the subject attended to the stimuli in the contralateral hemifield.  The
amplitude of modulation was about 20 percent of that evoked by alternating the
test stimuli every 18 sec with a uniform field.  Thus, neural activity in human
V1 is greater when a stimulus is relevant to a subject's task.  This is the
first evidence of attentional modulation in human area V1.
(Supported by NIH grants MH50228 and MH10897, Stanford University undergraduate
research opportunities grant, and Alfred P. Sloan fellowship.)


Heinrich H. Buelthoff
Max-Planck Institut fuer Biologische Kybernetik
Spemannstrasse 38
Tuebingen 72076, Germany

Title:  View-based Recognition

Brief Summary:  Theories of visual recognition may be divided into two major
classes: the view-based approach and the structural-description approach.  
View-based theories propose that recognition relies on features tied to the 
input image and therefore results in viewpoint dependence in the recognition of 
objects.  In contrast, structural description theories propose that recognition
relies on explicit specification of 3D (referred to as "Geons" by Biederman,
1987) and their spatial interrelations.  Object recognition, according to these
theories, is view invariant as long as these parts are visible.
  In the past we showed that recognition performance across a wide range of 
tasks and stimuli is viewpoint dependent and that qualitative differences in
shape do not typically result in viewpoint invariance.  Recently we started to
test recognition performance in large-scale structured environments. Scene
recognition is rarely studied because of the difficulty involved in isolation
and control of pertinent cues.  However, we can overcome such problems by using
computer graphics to model realistic 3D scenes and use VR technology to simulate
observer locomotion and interaction.  Our first experiments have attempted to 
determine whether there exists the same degree of view-dependency in scenes as
has been found for objects. We do this by using a single, sparsely decorated,
yet structured room with which subjects familiarize themselves. This learning
process can take three forms: active exploration, passive views, passive motion
displays ("backseat driver experiment").  In the active case, subjects can
maneuver in a restricted set of directions in order to find and acknowledge
"hidden" coded targets.  In the passive case, 2D views of the room are presented
to them in random sequence with some views containing embedded targets, which
they also have to acknowledge.  In the backseat driver experiment observers
passively viewed the movements of the "active-explorers."  Correct responses
and response latencies of observers in each condition were recorded in
subsequent (old/new) recognition tests.  Our results show:
(1)  After learning an environment only from one direction observers are also
able to recognize from novel vantage points but with higher error rates, similar
to object recognition.
(2)  The generalization to novel views is best after active learning.
(3)  Three-dimensional information from binocular stereo has no effect on 
recognition, similar to object recognition.
(4)  Active vision affects spatial encoding but the important factor is natural
locomotion through the scene and not volitional movement.
  These results provide further evidence for view-based encoding in humans and
further emphasize the importance of studying mental encoding and recognition
as a dynamic process. The generality of the view-based approach has also been
tested for recognition of dynamic objects and this work will be presented here
by Isabelle Buelthoff.
                                        In collaboration with Chris Christou.


Isabelle Buelthoff
Max-Planck Institut fuer Biologische Kybernetik
Spemannstrasse 38
Tuebingen 72076, Germany

Title:  Recognition of Dynamic Objects

Brief Summary:  We can often recognize objects not only on the basis of their
static appearance but also by observing how they move. The biological motion 
sequences devised by Johansson (1973) are elegant and powerful demonstrations of
this fact.  We tested whether observers exhibit the well-known canonical 
viewpoint effect while recognizing 3D biological motion sequences. Our results
show a remarkably impaired recognition performance with sequences recorded from
unusual viewpoints. This provides additional evidence for the role of viewpoint
familiarity and the inability of the visual system to extract view-independent 
representations.  To examine whether the motion traces used for recognition
preserve 3D information, or are largely 2D, we developed a special class of
biological motion sequences.  The distinguishing characteristic of these 
sequences was that while they preserve the "normal" 2D projections from one
viewpoint, their 3D structures were randomized.  Viewpoints preserving the
"normal" 2D projections yielded vivid biological motion percepts, whereas other
viewpoints yielded percepts of randomly moving dots.  In an additional set of
experiments we examined whether this result could be an outcome of a 
recognition-dependent top-down suppression of anomalies in 3D structures. Our
results indicate that subjects' expectations about 3D structure can suppress the
bottom-up depth information provided by binocular stereo.  Consistently we have
found that the extent of the top-down influence is reduced when the
recognizability of the objects is impaired by upside-down presentation.  Taken
together, these findings suggest that biological motion sequences are
represented by the human visual system as 2D traces rather than as 3D structural
descriptions, and that the perception of 3D structure may be based not only upon
low-level processes but also upon (**missing text**).


Jerome R. Busemeyer
Psychology Department
Indiana University
Bloomington In 47405-1301

Title:  Theoretical considerations for choice among diffusion models of response

Brief Summary:  There is a convergence of evidence pointing to the usefulness of
diffusion models for explaining response time phenomena. However, within this
class there are a variety of different subclasses (e.g., time invariant versus
time varying, linear versus nonlinear, ect.) Empirical evidence as well as
theoretical reasons for choosing a subclass will be reviewed. 


Thomas A. Busey
Department of Psychology
Indiana University
Bloomington, IN 47405

Title:  Summed-Similarity Accounts of Face Recognition

Summary: In four experiments we examined the false recognition of faces
within the context of a larger model of face recognition that accounts for
the effects of typicality and distinctiveness. Using morphing image processing
techniques, we created distracter faces that were mathematical blends of
two 'parent' faces. To address temporal context effects, during study the
two parent faces were seen either sequentially or separated by at least 20
other faces. During test, the two parent faces were replaced by a morph
distracter. We found very high false alarm rates to the morph distracters,
but no effects of the temporal context manipulation. In a forced choice
version, subjects were more likely to choose the distracter over the
parent face if the two parents are similar to each other, which is
consistent with a blending mechanism. Recognition models based on
Nosofsky's Generalized Context Model (GCM, Nosofsky, 1986) could account
for some but not all aspects of the data. A new model, called SimSample,
is developed that can account for the effects of typicality and
distinctiveness, but still has difficulty accounting for the high false
alarm rates to the morphs. A version that includes explicit prototype
representations can account for the morphs, as long as the prototype
strength is proportional to the similarity of the two parents. This mixed
model is consistent with a blending mechanism that is more likely to occur
between similar rather than dissimilar faces.


Asher Cohen
Psychology Dept.
Indiana University
Bloomington, IN 47405

Title:  Multiple Response Selection Systems

Brief Summary:  A widely held view states that response selection for all visual
stimuli is done by a central single mechanism.  I present evidence from three
different paradigms (time permitted) suggesting that there exist multiple
response selection mechanisms, some mediated by spatial attention and some not.
A model of spatial attention and response selection will be briefly described.


Matthew Dailey
CSE Dept. 0114
La Jolla, CA 92093-0114

Title: Is Face Processing "Special?":  Two Models and Future Directions

Brief Summary: There has been a great deal of progress, as well as controversy,
in understanding how complex objects, in particular human faces, are processed 
by the cortex.  At the same time, sophisticated neural network models have been
developed that do many of the same tasks required by these cortical areas.  Such
simplifying models allow us to explore hypotheses concerning relatively complex
domains such as face processing. In this talk, I will discuss two quite separate
models of face processing. One models how we perceive facial expressions, using
a quite general computational mechanism and architecture. The other uses another
quite general technique to suggest how face processing might develop into a
"module". Both give rise to testable hypotheses about how face processing is
actually done.


Ruth S. Day
Psychology Department
Duke University
Box 90085
Durham, NC 27708-0085

Title:  The Model Penal Code:  Alternative Representations and Cognitive

Brief Summary:  The Model Penal Code is used widely in criminal cases.  One
section provides rules for determining whether a defendant is "not guilty by
reason of insanity."  Although it contains only 44 words, juries (and many legal
experts) have a notoriously difficult time understanding it and applying it to
court cases.  In a series of experiments, participants studied this excerpt
along with an addendum designed to emphasize the key criteria it contains; the
addendum was provided in either list or tree format.  Then partipants tried to
decide whether various case scenarios met Code requirements.  The alternative 
representations had cognitive consequences -- they affected comprehension,
memory, and decision making.  Other conceptual, linguistic, and procedural
factors were also involved.  This work holds implications for understanding
complex cognition in general, as well as how to improve jury deliberation and
decision making.


David Diller
Department of Psychology
Indiana University
Bloomington, IN 47405

Title: Information Processing From Unattended Visual Locations

Brief Summary: George: Rich Shifrin, Asher Cohen & I have quite a few studies to
report regarding whether information is processed from a visual location during
the time that attention is fully directed to another visual location. Using a
primary RSVP task, we narrowly focus attention at a central location, then
occassionally we interrupt this task with a secondary task to assess unattended
processing.  Until recently, we'd found very little evidence for unattended
processing using a wide variety of secondary tasks.  Recently using a word-
fragment completion paradigm we seem to have evidence for automatic processing.
Our studies suggest processing occurs to the level of visual features, and quite
possibly to the lexical level or even to the level of meaning.  We are currently
running additional studies, that will be presented @ AIC, to 1) ensure ourselves
that attention is fully allocated to the central location & the primary task and
2) to assess the depth of the processing that occurs without attention.


David Heeger 
Department of Psychology
Stanford University
Stanford, CA 94305

Title: Brain activity in visual cortex predicts reading performance in dyslexia.

Brief Summary:  The relationship between brain activity and reading performance
was examined to test the hypothesis that dyslexia involves a deficit in a 
specific visual pathway known as the magnocellular (M) pathway.  Functional
magnetic resonance imaging (fMRI) was used to measure brain activity in dyslexic
and control subjects in conditions designed to preferentially stimulate the M
pathway.  Dyslexics showed reduced activity compared to controls both in primary
visual cortex (V1) and in a secondary cortical visual area (MT+) that is
believed to receive a strong M pathway input.  Most importantly, significant
correlations were found between individual differences in reading rate and brain
activity.  These results support the hypothesis for an M pathway abnormality in
dyslexia and imply a strong relationship between the integrity of the M pathway
and reading ability.


Lenny Kontsevich
Smith-Kettlewell Eye Research Institute
2232 Webster Street
San Francisco, CA 94115

Title: Sensory and Cognitive Components of the Position Coding

Brief Summary: I will present new striking data indicating that position
judgement is based on very few samples of the image. Then I'll describe a new
model that incorporates the new findings and explains the major regularities
known for the position coding.


Dan Levin
Psychology Department
Kent State University
Kent, OH 44242

Title:  Combining views in natural scenes:  Perception and metaperception in
motion pictures and everyday interactions

Summary: In this talk I will discuss intuitions that filmmakers have
regarding the psychology of scene perception, and will describe research by
Dan Simons and myself that embodies many of these intuitions. I will also
discuss more recent work that directly tests filmmakers' hypotheses
regarding the centrality of specific cues such as gaze direction in
specifying the relative location of novel views. Generally, the talk will
emphasize the importance of understanding not only perceptual and cognitive
processes in scene perception, but also the metacognitive issues invoked by
subjects apparent misunderstanding of these processes.


Geoffrey R. Loftus
Department of Psychology                        
University of Washington                        
Seattle, WA 98195                               

Title: Models of the Relation Between Confidence an d Accuracy.

Brief Summary:  Confidence and accuracy, while often considered to tap the 
same memory representation, are often found to be only weakly correlated (e.g.
Deffenbacher, 1980; Bothwell, Deffenbacher & Brigham, 1987). There are two
possible (nonexclusive) reasons for this weak relation. First, it may be simply
due to noise of one sort or another; that is, it may come about because of 
both within- and between-subject statistical variations that are partially
uncorrelated for confidence measures on the one hand and accuracy measures on
the other. Second, confidence and accuracy may be uncorrelated because they are
based, at least in part, on different memory representations that are affected 
in different ways by different independent variables. In this talk, I will
propose a general theory that is designed to encompass both of these
possibilities and, within the context of this theory, evaluate effects of four
variables:  degree of rehearsal, study duration, study luminance, and test
luminance in three face-recognition experiments. In conjunction with the theory,
the results allow us to begin to identify the circumstances under which
confidence and accuracy are based on the same versus on different sources of
information in memory. In particular, we conclude the following. First,
prospective confidence (assessed at the time of original study) and eventual
accuracy are based on different sources of information: A sufficient description
of these differences is that accuracy is determined by what we term memory
strength, while prospective confidence is based both on memory strength and on
a second dimension which we term memory certainty. Second, given identical test
circumstances, retrospective confidence (assessed at the time of test) and
accuracy can be considered to be based on the same source of information, memory
strength. Third, degrading a picture at test can be construed as differentially
affecting strength and certainty which, in turn, produces different effects on
confidence and accuracy.


Zhong-Lin Lu
Department of Psychology
University of Southern California
Los Angeles, CA 90089-1061

Title:  Brain-wave Recognition of Words

Authors:  Zhong-Lin Lu, Patrick Suppes and Bing Han
Brief Summary:  Electrical (EEG) and magnetic (MEG) brain waves of seven
subjects under three experimental conditions were recorded for the purpose of
recognizing which one of seven words was processed.  The analysis consisted of
averaging over trials to create prototypes and test samples, to both of which
Fourier transforms were applied, followed by filtering and an inverse
transformation to the time domain. The filters used were optimal predictive
filters, selected for each subject and condition.  Recognition rates, based on
a least squares criterion, varied widely, but all but one of 24 were
significantly different from chance.  The two best were above 90 percent.  These
results show that brain waves carry substantial information about the word being
processed under experimental conditions of conscious awareness.


Larry O'Keefe
Center for Neural Science
New York University
4 Washington Place, Rm 809
New York, NY 10003

Title: Response variability in cortical neurons

Brief Summary:  I'll talk about how "noisy" signals from single neurons are, and
where the noise might arise from.  I'll show data from MT neurons in alert and
anesthetized monkeys, and show the influences of eye movements and stimulus
properties affect response variability.  I'll try to place this in a context of
understanding how neuronal signals might be pooled/combined to yield perceptual

Long Summary:

L. P. O'Keefe*, W. Bair and J. A. Movshon. Center for Neural Science
and Howard Hughes Medical Institute, New York University, New York 10003.

The performance of sensory systems is limited by the variability of neuronal
activity. Studies of neurons in visual cortex of anesthetized cat and monkey and
alert monkey show that response variance is related to mean response by a power
function with an exponent >= 1. Anecdotal claims exist that variability is lower
in alert animals than in anesthetized ones, and it has recently been reported
that the variability of V1 neurons in the alert macaque is lowest in periods
during which gaze is most stable, suggesting that fixational eye movements are
an important component of response variability in alert animals (Gur et al.,
J. Neurosci, 17: 2914-2920, 1997).  We have now compared response variability in
macaque MT neurons recorded from paralyzed opiate-anesthetized animals and from
alert animals trained to perform a fixation task, using the same moving visual
stimuli in both preparations. We also examined the effect of fixational eye
movements by analyzing response variability during periods of stable gaze. In
both preparations, response variance was approximately proportional to mean
response. The power-law exponent was modestly greater for data from anesthetized
macaques. Variability of response measured over entire trials in the alert
monkey was indistinguishable from that measured during periods of stable gaze. 
We conclude that the factors determining the variability of MT responses are
largely intrinsic to the visual system, and are not substantially different
under anesthesia. More specific analyses will reveal whether these small
differences are important and will allow us to determine how measurements made
in different preparations may be meaningfully compared to assess the role of
neuronal variability in limiting sensory performance.


Alice J. O'Toole
School of Human Development, GR 4.1
University of Texas at Dallas
Richardson, TX 75083-0688

Title:  Three-dimensional Caricatures of Human Faces:  As we get older, do we
get more distinct?

Brief Summary:  We applied a standard facial caricaturing algorithm to a three-
dimensional representation of human heads.  This algorithm produced an increase
in the apparent age of the face --- both at a local level, by exaggerating small
facial creases into wrinkles, and at a more global level via changes that seemed
to make the underlying structure of the skull more evident.  We discuss these
results in terms of the importance of the nature of the features made more
distinct by a caricaturing algorithm and the nature of human representation(s)
of faces.


John Palmer
Psychology Box 351525
University of Washington
Seattle, WA 98195

Title: Is visual search consistent with high threshold or signal detection

Brief Summary:  Theories of visual search are typically based upon either high
threshold or signal detection theory.  These alternative theories were compared
in a variety of search accuracy experiments including manipulations of set size,
number of targets, stimulus discriminability, response bias, external noise, and
distractor heterogeneity.  In all csases allowing a test, results from search
accuracy experiments were consistent with the predictions of signal detection
theory and were inconsistrent with high threshold theory.


Allen Poirson
Psychology Dept., Bldg. 420
Stanford University
Stanford, CA 94305

Title:  Color tuning to moving stimuli in the human visual cortex measured
using fMRI

Brief Summary: I evaluate color processing in an area of the human brain that is
highly responsive to motion using functional magnetic resonance imaging (fMRI).
This area, which I informally call the M-complex, has been studied by a number
of groups, and is thought to be to a collection of motion areas homologous to
those found near macaque area MT.  The specific focus of the talk will be how
signals initiated in the Short Wavelength Cones (or blue cones) drive the fMRI
signal in the M-complex.
  I address this issue for three reasons.  First, a series of anatomical and
physiological papers have demonstrated that there is a specialized S-cone
pathway within the retina.  Recently this specialized pathway has been shown
to continue into the middle brain.  Where does the S-cone pathway go once it
reaches the cortex?  Second, measurements of neuron's color tuning in monkey MT
show no S-cone input to monkey area MT.  Are humans like monkeys?  To the extent
that M-complex activity parallels monkey MT activity, we should find no activity
based on S-cone initiated signals.  Third, psychophysicists have noted that
visual stimuli that excite only the S-cones appear to move more slowly than
equivalent physical stimuli that excite the other two cone types.  These motion-
color illusions can be very strong so that in some cases a moving stimulus
appears to stop altogether.  To the extent that activity in the M-Complex
correlates closely with the visual percept of motion, as is sometimes claimed,
we should see little M-complex activity whenever there is a weak motion percept. 
When the motion percept returns, so should the M-complex response.
  I conclude that the human M-complex receives input from all three classes of
photoreceptors, including the S-cones.  The activity I observe in the M-complex
is not simply connected to the perception of motion.


Bill Prinzmetal
Psychology Department
University of California
Berkeley, CA 94707 

Title:  From the Ponzo to the Poggendorf Illusions: The Mystery Spot Reveals
Its Secrets 

Brief Summary:  We will present a unified theory to explain the Ponzo Illusion,
the Poggendorf Illusion, the Zollner Illusion, and other visual illusions.  The
theory is based on the distorted room illusion, an example of which is the
Mystery Spot roadside attraction in Santa Cruz, CA.  In a room exhibiting these
deceptions, water appears to flow uphill, people appear capable of standng on
on walls, and the apparent height of objects is dramatically altered.  A general
theory of the distorted room illusion is that the contours of the room cause a
misperception of vertical and/or horizontal, as in the rod and frame effect.
  We will illustrate that the same factors that cause the distorted room 
illusion are capable of explaining a wide variety of other visual illusions.
Furthermore, we will compare our explanation of these illusions with previously
presented hypotheses such as (1) inappropriate application of depth cues; 
(2) low spatial frequency band pass filtering; (3) interactions between
orientation tuned neural units; (4) subterranean vortex generators such as
might be thought to cause the Bermuda Triangle.


Roger Ratcliff
102 Swift Hall
Northwestern University
2029 Sheridan Road
Evanston, IL 60208

Title:  A Diffusion Model Account of Masking in Two-choice Letter Identification

Authors:  Roger Ratcliff and Jeffrey N. Rouder
Brief Summary:  The diffusion model of Ratcliff (1978, 1981, 1988) is applied to
data from two letter identification experiments.  The letters are masked and SOA
is manipulated to vary response probability from near chance to near ceiling. 
In a block of trials, only two letters are presented and subjects are required
to press one of two keys to indicate which letter was presented.  One experiment
uses a free response procedure and reaction time and accuracy are the
experimental data to be fitted.  The second experiment uses a deadline procedure
and the growth of accuracy as a function of time is fitted.  Two views of
masking are contrasted, one in which the output of perception to the decision
process is time varying so the drift rate in the decision process rises then
falls after the mask is presented.  The second view is that drift rate is 
constant after stimulus presentation and reflects integration of the stimulus.
The data are fit by the diffusion model with constant drift (including reaction
time distributions and the relative speeds of correct and error responses) and
various parameter invariances are demonstrated across conditions.


Richard M. Shiffrin
Psychology Department
Indiana University
Bloomington, IN 47405

Title:  A New Model for Recognition Response Times

Brief Summary:  Typical response time models have the subject monitor continu-
ously the current evidence for and against the test item being old or new, re-
sponding when the evidence reaches a criterion level. Peter Nobel collected data
arguing against such a model in a recognition memory setting.  David Diller and 
I developed a model in which the subject monitors instead the amount of evidence
accumulated, rather than the evidence for and against, interrupting the process
& reading out the results when sufficient evidence has been taken into account.
Combined with an assumption that forgetting occurs during the course of
retrieval, this model provides an excellent account of the results.


Daniel J. Simons
Psychology Department
Harvard University
33 Kirkland St., Room 820
Cambridge, MA  02138

Title:  Representing spatial layout across changes in view

Brief Summary:  Recent studies of change blindness show that relatively little
visual information is retained and used from one view of a scene to the next.
However, people often can detect changes to the spatial relations among objects
suggesting that such representations are preserved across views and may underlie
our experience of a stable visual world.  In this talk I will describe a recent
series of experiments on representations of spatial layout that examine (a) the
stability of layout representations across views and (b) differences between the
effects of observer movement and display rotation on layout change detection. 
The findings reveal a potentially important disparity between view changes
caused by rotating a visual display (orientation changes) and view changes
caused by shifts in the observer's viewing position (viewpoint changes).
Specifically, orientation changes disrupt change detection, but viewpoint 
changes of equivalent magnitude do not.  The mechanisms underlying this 
difference in performance will be discussed.


Dirk Smit
Division of Psychology
Staffordshire University
College Road   
England ST4 2DE

Title: On the additivity of visual grouping by proximity, similarity, and good
continuation in dot lattices.

Brief Summary: 	We tried to quantify the Gestalt law of grouping by proximity.
This resulted in a simple model which only takes the distance between dots into
account. We next investigated the effect of similarity and good continuation and
found that the initial model stayed intact. We concluded that the effects of
proximity, similarity and good continuation on perceptual grouping are additive.


Philip L. Smith
Department of Psychology
University of Melbourne
Parkville, Vic. 3052

Title:  A Dynamic Stochastic Theory of Visual Attention

Brief Summary: This paper describes a new theory of visual attention which seeks
to characterise the dynamic effects of selective attention on low-level visual
detection and discrimination. The theory is an amalgam of two existing theories:
the multichannel, leaky stochastic integration (MLSI) theory of visual detection
of Smith (Psych. Rev., 1995) and the attention gating theory (AGT) of Sperling &
colleages (e.g., Sperling & Weichselgartner, Psych. Rev., 1995).  AGT proposes
that the effects of attention can be parsed into a sequence of discrete atten-
tional episodes, each with a quantifiable temporal and spatial extent.  MSLI
theory provides a way to model dynamic variations in the uptake of information
from a display during the course of an experimental trial. When combined, these
2 approaches provide a way to model performance in divided attention and spatial
cuing paradigms.  The resulting theory yields detailed predictions about the ef-
fects of attention on response latency and accuracy at the level of the reaction
time distribution and at the level of the psychometric function, respectively.


Update on the Three Systems Theory of Motion Perception.

George Sperling and Zhong-Lin Lu
U. of California, Irvine, and U. of Southern California, Los Angeles, CA

Abstract: Recent psychophysical innovations suggest the presence of three
systems of perceptual motion computation.  The methods utilize complex
computer-generated motion stimuli in conjunction with combinations of six
paradigms (below).
  The three motion systems: First-order detects movement of (ordinary)
luminance modulations.  Second-order can detect motion in stimuli in which
expected luminance is the same everywhere but some property, such as an area
of higher contrast or of flicker moves; third-order detects movement of a
salience marker, that is, changes in location of areas marked as "figure"
or simply as "important." Both first- and second-order mechanisms use a
primitive motion-energy algorithm, that is almost exclusively monocular and
equally fast (cutoff freq approx 10-12 Hz).  They compute motion independently
of each other.  The third-order mechanism may use the same algortithm but it's
input is very coarsely quantized.  The third-order mechanism is binocular,
(i.e., indifferent to the eye of origin), slow (cutoff freq around 3-4 Hz),
but extremely versatile.  It acts on bottom up information computing motion
from luminance-modulation, texture-contrast modulation, depth-modulation,
motion-modulation, flicker-modulation, and from other types of stimuli but,
unlike first- and second-order systems, it is subject to top-down\-\-e.g., 
control via attention.  Attentional manipulations can generate apparent motion
in displays where none would be seen otherwise, and attention can determine
the direction of apparent motion in ambiguous displays.
  Six paradigms: (1) measuring temporal (and spatial) tuning functions: first
and second-order are virtually identical, third order is 3x slower; (2) masking
by pedestals (i.e., masking of moving sinewave gratings by stationary gratings
of the same spatial frequency): first- and second-order are unaffected by
2x pedestals whereas third-order is destroyed, second-order (only) is resistant
to pedestals of any size; (3) interocular presentations: third-order is
resistant, first- and second-order fail; (4) stimulus superpositions with
varying phases: phase independence for first- and seocnd order indicates
separate mechanisms; (5) selective adapatation: each system can be selectively
adapted independently; (6) attentional manipulations affect only third-order;
and (7) selective effects of localized brain lesions: only selective lesions
for first- and second-order have been discovered so far.
  Separate left- and right-eye computations for first-order and second-order
imply a total of five perceptual motion computations, all carried out
concurrently.  Together, the system properties and relations define a
functional architecture of human visual motion perception.  As time permits,
we will consider how to generate (almost) pure stimuli for each system, the
evidence for various of the points made above, and the interpretation of
various problematic paradigms (such as second-order reverse phi).


Jim Tanaka
Department of Psychology
Oberlin College
Oberlin, OH 44074

Title:  Attractor Fields in Face Recognition

Brief Summary:  Previous research has shown that a familiar face can be 
recognized across many changes in the stimulus input.  The many-to-one mapping
of stimulus inputs to a single face memory is referred to as a face
representation's attractor field.  According to the density hypothesis, the span
of an attractor field is affected by the density of nearby representations in a
multidimensional "face space."  In this research, the density hypothesis was
tested by asking subjects to make likeness judgments to morphed face images that
varied in their contribution of atypical and typical "parent" faces.  Consistent
with the density hypothesis, face images containing a 50/50 contribution from an
atypical and typical parent face were judged as bearing a stronger resemblance
to the atypical face parent than the typical face parent in immediate memory and
direct perception tasks.  The computational basis of the atypicality bias was
demonstrated in a neural network simulation where morph inputs of atypical and
typical representations elicited stronger activation of atypical output units
than the typical output units.  Together, the behavioral and simulation evidence
support the central prediction of the density hypothesis; specifically, the
attractor fields of atypical faces span over a broader region of face space than
the attractor fields of typical faces.  


Michael J. Wenger
Department of Psychology
Social Sciences 2
UC Santa Cruz
Santa Cruz, CA 95064

James T. Townsend
Department of Psychology
Indiana University 
Bloomington, IN  47405-6401

Title:  Information processing and facial processing: Old challenges and new
(theoretical) technology

Brief Summary:  Many of the questions that have been and continue to be of crucial
concern in understanding facial cognition are questions that have a long history
in the study of cognition in general. And many of these questions, such as
whether two aspects of a stimulus (e.g., the eyes and the mouth) are processed
sequentially or concurrently, or whether there might be limitations in a
system's capacity to process those aspects, have appeared to be extremely
difficult to address.  However, much progress has been made in producing general
mathematical characterizations of these problems and experimental approaches
that allow those characterizations to be tested. In this talk, we present a
review of some of these advances and the ways in which we are currently applying
them to issues in facial cognition.


Samuel J. Williamson
Department of Physics
New York University
4 Washington Place
New York, NY 10003-1113

Title:  Neuromagnetic fields reveal cortical plasticity when learning an
auditory discrimination task

Authors:  S.J. Williamson and S. Cansino
Brief Summary:  Auditory evoked neuromagnetic fields were recorded while
subjects learned to discriminate small differences in frequency and intensity
between two consecutive tones. When discrimination was no better than chance,
evoked field patterns across the scalp manifested no significant differences
between correct and incorrect responses.  However, when performance was correct
on at least 75% of the trials, the field pattern differed significantly between
correct and incorrect responses during the first 70 ms following the onset of
the second tone.  In this respect, the magnetic field pattern predicted when the
subject would make an incorrect judgment more than 100 ms prior to indicating
the judgement by a button press. 


George Wolberg
Dept. of Computer Science
City College of New York
138th St. at Convent Ave., Rm. R8/206
New York, NY 10031

Title:  Hierarchical Image Registration for Affine Motion Recovery

Brief Summary: This talk will describe a hierarchical image registration
algorithm for affine motion recovery. The algorithm estimates the affine
transformation parameters necessary to register any two digital images
misaligned due to rotation, scale, shear, and translation. The parameters are
computed iteratively in a coarse-to-fine hierarchical framework using a
variation of the Levenberg-Marquadt nonlinear least squares optimization method.
This approach yields a robust solution that precisely registers images with
subpixel accuracy.  We demonstrate the algorithm on pairs of digital images
subjected to large motion.


Sophie Wuerger
Communication & Neuroscience
Keele University
Keele ST5 5BG
United Kingdom

Title: 	Blur tolerance in various directions in colour space

Brief Summary: 	We investigated how much blur the visual system tolerates in
various directions in colour space. We measured blur thresholds by convolving a
square wave grating with gaussian masks of varying standard deviations. For
stimuli of equal cone contrast we found that the same amount of blur was
tolerated in the red-green and the luminance direction (standard deviation of
gaussian = 1 minute of visual angle). This result is consistent with the idea
that the red-green and the luminance pathway have similar spatial properties. In
the yellow-blue directions we found much higher blur thresholds (2-8 minutes of
visual angle). A better understanding of the spatio-chromatic properties of the
visual system might be helpful in designing more efficient image compression


Last revised: 29jan98