Attention, Perception, and Temporal Context
David Pascucci, École Polytechnique Fédérale
de Lausanne, Switzerland.
Árni Kristjánsson, University of Iceland
& National Research University Higher School of Economics, Moscow, Russia.
Introduction
It is our pleasure to introduce a three-day
online workshop “Attention, Perception and Temporal Context”, taking
place from January 13th to 15th, 2021.
Recent years have seen tremendous progress in
the scientific understanding of the role of prior experience and temporal
context in human perception and cognition. The workshop will highlight advances
from many of the key researchers in the field who have contributed to the constellation
of novel findings on attentional history, perceptual serial dependence, and
statistical learning. We have
attempted to bring together some of the leading frameworks in this field, to
promote new empirical and theoretical links between the effects of previous
events on attention and perceptual processing.
Speaker: Title:
Speaker: Title:
Speaker: Title:
|
Niko Gekas |
Manipulation of stimulus contrast reveals its
complex role in perceptual aftereffects |
|
Jan Brascamp |
Attractive and repulsive history effects in visual perception |
|
David Pascucci |
Guido Marco Cicchini
Institute
of Neuroscience, National Research Council, Pisa, Italy
One
function of perceptual systems is to construct and maintain a reliable
representation of the environment. This strategy is to cope with noisy information, but can lead to stimuli being biased towards perceptual
history, a phenomenon known as serial dependence. It is still unclear whether
serial dependence biases sensory encoding, or only perceptual decisions. We
leveraged on the “surround tilt illusion” – where tilted flanking stimuli
strongly bias perceived orientation – to measure its influence on the pattern
of serial dependence. When a neutral stimulus preceded by an illusory one
maximal serial dependence occurred when the perceived orientations matched,
suggesting that the sensory history incorporates contextual biases. However,
when an illusory stimulus was preceded by a neutral stimulus, maximal serial dependence
occurred when their physical orientations matched, suggesting that sensory
history interacts with incoming sensory signals before they are biased by
flanking stimuli. The evidence suggests that priors are high-level constructs
incorporating contextual information, which interact directly with early
sensory signals, not with highly processed perceptual representations. It also
suggests that serial dependences arise at a network rather than local level.
Cora Fischer
Institute of Medical Psychology, Goethe-University, Frankfurt, Germany
Visual perception operates in an object-based
manner, by integrating associated features via attention. Working memory allows
a flexible access to a limited number of currently relevant objects, even when
they are physically no longer present. Recently, it has been shown that we
compensate for small changes in an object’s appearance over memory episodes to
ensure its perceived continuity. This phenomenon was termed 'serial dependence’
and has mostly been studied in situations that comprised only a single relevant
object. I will show that in situations with multiple objects, context features
like color, temporal or spatial position are used as anchors to selectively
integrate corresponding objects over memory episodes. Moreover, I will present preliminary
results based on the high temporal resolution of the MEG signal suggesting that
serial dependence operates during access to object information in working
memory rather than during the object’s perception or maintenance.
Albert Compte
Institut
d’Investigacions Biomèdiques
August Pi i Sunyer
(IDIBAPS), Barcelona, Spain
Working memory plays a central role in cognition, and yet its
fundamental neural mechanisms remain a matter of debate. Persistent activity
has been long considered, but recently subthreshold mechanisms have been
proposed as alternative. Here, we show how the interaction of both mechanisms
determines serial biases of monkeys and humans in spatial working memory.
Stimulus information disappears from electrophysiological signals between trials, but remains present in the synchrony of prefrontal
neurons as predicted by a computational model integrating attractor dynamics
and short-term plasticity. Just prior to the new stimulus, this
subthreshold trace in some trials is reignited into activity that recapitulates
the previous stimulus representation. We show that the reactivation strength
correlates with the strength of serial biases in both monkeys and humans.
Single-pulse TMS applied to the prefrontal cortex prior to trial start enhanced
serial biases, demonstrating the causal role of prefrontal reactivations in
determining serial dependence in working memory.
Dominique
Lamy
Department of Psychology, School of Psychological
Sciences, Tel Aviv University, Israel
Where we shift our
attention next is thought to be the location with the strongest combined
representation of salience and relevance on the “priority map” at any given
moment. But are we at the mercy of the constant changes occurring in our
environment, and automatically move our attention to the ever-changing location
with the highest priority? Or do we wait for clues that the appropriate moment
has arrived to deploy our attention? Here, we address this hitherto neglected
issue using a spatial-cueing paradigm. We examined whether attention is
deployed as soon as a salient change occurs (the cue), or only when the context
signaling that attention should be deployed appears (the search display). The
findings from three experiments relying on manual responses and eye movements
support the latter possibility. These findings challenge traditional views on
how attentional deployment is triggered. They are discussed within the Priority
Accumulation Framework (PAF).
Omer Daglar Tanrikulu
Vision Sciences Laboratory, School of
Health Sciences, University of Iceland, Reykjavik, Iceland.
Priming of attention shifts have recently been used as an effective
method to uncover internal representations of visual ensembles. This method,
Feature Distribution Learning (FDL), takes advantage of attentional priming
when role reversals between target and distractors occur in visual search.
Results obtained with the FDL method have shown that the visual system can
encode feature probability distributions in detail and shows sensitivity to
temporal perturbations in the statistical properties of visual scenes. In this
talk, I will introduce the FDL method and review studies utilizing it. I will
then present our latest study investigating the temporal dynamics of the FDL
method, in which we investigated how observers temporally integrate two
different orientation distributions from sequentially presented visual search
trials. The robust recency effects obtained in this study indicated that the
visual system prefers to utilize sensory history when the statistical
properties of the environment are relatively stable.
Heleen A. Slagter
Vrije Universiteit
Amsterdam, Amsterdam, The Netherlands
Perception is more than meet’s
the eye; how we experience the world is critically shaped by attention (what is
relevant) and as a rapidly growing body of work indicates, by predictions
grounded in past experience and regularities in the outside world (what is
likely). Overturning the classical notion of perception as a largely bottom-up
process, the idea that our brain is a prediction machine, continually trying to
predict what is ‘out there’ based on the probabilistic structure of the world,
is quickly growing in scientific stature and influence. Yet, little is still
known about how predictions, independently or in interaction with attention,
may shape perception and performance. In my talk, I will present findings from
several behavioral and EEG studies that examined how predictions are neurally implemented and guide behavior, and how this may
depend on the goal-relevance (target, distractor), modality (visual, auditory),
and/or dimension (spatial, nonspatial, temporal) of the predicted information. Collectively,
these findings reveal how the brain rapidly picks up on the probabilistic
structure of the world, guiding what we attend to and what we ignore.
Nikos Gekas
School
of Psychology, University of Nottingham, Nottingham, UK
Visual perception is a perpetually adapting
process where the current percept is affected by what has been previously
observed. It has been shown that stimuli seen at different times in the past
can have opposite effects on the perceived orientation of the current stimulus.
Recent history has a repulsive effect on perceived orientation, while more
distant history appears to have an attractive effect. Here, we test how the
short-term repulsion and long-term attraction are modulated by the saliency of
the stimuli presented in the past history. The contrast of both past and current
stimuli were simultaneously manipulated in such a way as to decrease the
repulsive effect and increase the attractive effect, or vice versa. We found
that the manipulation of contrast had a strong effect on the perceptual
aftereffects in our orientation discrimination task, which suggests a complex
relation between contrast, orientation, and adaptation.
Jan W. Brascamp
Department
of Psychology, Michigan State University, US
In what way is perception of current stimuli
altered by what was seen a short while ago? A first distinction is between
attractive and repulsive effects. Attractive effects are when the present
stimulus appears more similar to a recent stimulus, or when a present stimulus
that physically resembles a recent stimulus is more easily detectable.
Repulsive effects are the opposite: the present stimulus looks more dissimilar
to a previous stimulus than it actually is, or is
harder to detect if it physically resembles the previous stimulus. Although
both types of history effects are well known, it is surprisingly unclear what
factors determine which of these two diametrically opposite effects arises in
any given situation. I will discuss my work on bistable perception, where both
kinds of effects are observed: in some cases observers tend to perceive an
ambiguous stimulus in a way that matches prior experience; in other cases they
tend to perceive the opposite interpretation. I will also review a set of
results from outside that field. Based on data patterns observed across fields
I will tentatively suggest as one important factor in this context a balance
between processing at early stages versus advanced stages of the visual
hierarchy. In this suggested scheme prior stimuli that are associated with a
stronger bottom-up drive and a pronounced representation at early processing stages
tend to leave a repulsive trace, whereas relatively weak stimuli whose
processing becomes more pronounced at later hierarchical stages (for instance
due to perceptual dominance in bistable perception) tend to lead to attraction.
Aside from offering this scheme, I will also speculate about the functional
significance of these two opposite types of history effects.
David Pascucci
Laboratory of Psychophysics, Ecole Polytechnique
Fédérale de Lausanne (EPFL), Switzerland.
Human perception and perceptual decisions are
highly influenced by their temporal context: There is almost no elementary percept
that can be isolated from its preceding history. In psychophysical tasks, this
leads to sequential effects in perceptual decisions, in which the judgment of a
given stimulus feature is affected by the history of preceding trials before.
Such effects can be attractive (assimilative) when features of stimuli are
erroneously reported as similar to previous trials, or repulsive (contrastive)
when the difference between the present and the past stimulus is exaggerated.
These opposite biases can be observed under a variety of conditions,
complicating a unified understanding. In this talk, I will review some of my
recent work on the assimilative effect known as serial dependence (SD),
in an attempt to elucidate its nature. I will present evidence that, in
perceptual tasks, SD occurs beyond basic visual features and perceptual objects
and I will describe cases in which contrasting effects dominate. I will propose
that SD resides at an abstract and task-dependent level of representation,
where the selectivity to features and objects is lost. The perspective I will
outline is not a conclusive framework, but a starting point for debate and
discussion towards a cohesive view.