 |
Member-Initiated Symposia2012 Symposia Distinguishing perceptual shifts from response biases Human visual cortex: from receptive fields to maps to clusters to perception Neuromodulation of Visual Perception Part-whole relationships in visual cortex Pulvinar and Vision: New insights into circuitry and function What does fMRI tell us about brain homologies?
|
Distinguishing perceptual shifts from response biasesFriday, May 11, 3:30 - 5:30 pm Organizer: Joshua Solomon, City University London Presenters: Sam Ling, Vanderbilt;
Keith Schneider, York University; Steven
Hillyard, UCSD; Donald MacLeod, UCSD;
Michael Morgan, City University London, Max Planck Institute
for Neurological Research, Cologne; Mark Georgeson,
Aston University Symposium DescriptionSensory adaptation was originally considered a low-level phenomenon
involving measurable changes in sensitivity, but has been extended to include
many cases where a change in sensitivity has yet to be demonstrated. Examples
include adaptation to blur, temporal duration and face identity. It has
also been claimed that adaptation can be affected by attention to the adapting
stimulus, and even that adaptation can be caused by imaging the adapting
stimulus. The typical method of measurement in such studies involves a
shift in the mean (p50) point of a psychometric function, obtained by the
Method of Single Stimuli. In Signal Detection Theory, the mean is
determined by a decision rule, as opposed to the slope which is set by internal
noise. The question that arises is how we can distinguish shifts in mean due to
a genuine adaptation process from shifts due to a change in the observer’s
decision rule. This was a hot topic in the 60’s, for example, in
discussion between Restle and Helson over Adaptation Level Theory, but it has
become neglected, with the result that any shift in the mean of a psychometric
function is now accepted as evidence for a perceptual shift. We think
that it is time to revive this issue, given the theoretical importance of
claims about adaptation being affected by imagination and attention, and the
links that are claimed with functional brain imaging. Presentations Attention alters appearance
Sam Ling, Vanderbilt Maintaining veridicality seems to be of relatively low priority for
the human brain; starting at the retina, our neural representations of the
physical world undergo dramatic transformations, often forgoing an accurate
depiction of the world in favor of augmented signals that are more optimal for
the task at hand. Indeed, visual attention has been suggested to play a key
role in this process, boosting the neural representations of attended stimuli,
and attenuating responses to ignored stimuli. What, however, are the
phenomenological consequences of attentional modulation? I will discuss a
series of studies that we and others have conducted, all converging on the
notion that attention can actually change the visual appearance of attended
stimuli across a variety of perceptual domains, such as contrast, spatial
frequency, and color. These studies reveal that visual attention not only
changes our neural representations, but that it can actually affect what we
think we see. Attention increases salience and biases decisions but does not alter
appearance.
Keith Schneider, York University Attention enhances our perceptual abilities and increases neural
activity. Still debated is whether an attended object, given its higher
salience and more robust representation, actually looks any different than an
otherwise identical but unattended object. One might expect that this
question could be easily answered by an experiment in which an observer is
presented two stimuli differing along one dimension, contrast for example, to
one of which attention has been directed, and must report which stimulus has
the higher apparent contrast. The problem with this sort of comparative
judgment is that in the most informative case, that in which the two stimuli
are equal, the observer is also maximally uncertain and therefore most
susceptible to extraneous influence. An intelligent observer might
report, all other things being equal, that the stimulus about which he or she
has more information is the one with higher contrast. (And it doesn't
help to ask which stimulus has the lower contrast, because then the observer
might just report the less informed stimulus!) In this way, attention can
bias the decision mechanism and confound the experiment such that it is not
possible for the experimenter to differentiate this bias from an actual change
in appearance. It has been over ten years since I proposed a solution to
this dilemma—an equality judgment task in which observers report whether the
two stimuli are equal in appearance or not. This paradigm has been
supported in the literature and has withstood criticisms. Here I will
review these findings. Electrophysiological Studies of the Locus of Perceptual Bias
Steven Hillyard, UCSD The question of whether attention makes sensory impressions appear
more intense has been a matter of debate for over a century. Recent
psychophysical studies have reported that attention increases the apparent
contrast of visual stimuli, but there is still a controversy as to whether this
effect is due to the biasing of decisions as opposed to the altering of
perceptual representations and changes in subjective appearance. We
obtained converging neurophysiological evidence while observers judged the
relative contrast of Gabor patch targets presented simultaneously to the left
and right visual fields following a lateralized cue (auditory or visual).
This non-predictive cueing boosted the apparent contrast of the Gabor target on
the cued side in association with an enlarged neural response in the
contralateral visual cortex that began within 100 ms after target onset.
The magnitude of the enhanced neural response in ventral extrastriate visual
cortex was positively correlated with perceptual reports of the cued-side
target being higher in contrast. These results suggest that attention
increases the perceived contrast of visual stimuli by boosting early sensory
processing in the visual cortex. Adaptive sensitivity regulation in detection and appearance
Donald MacLeod, UCSD The visual system adapts to changing levels of stimulation with
alterations of sensitivity that are expressed both in in changes in
detectability, and in changes of appearance. The connection between these two
aspects of sensitivity regulation is often taken for granted but need not be
simple. Even the proportionality between 'thresholds' obtained by self-setting
and threshold based on reliability of detection (e.g. forced-choice) is not
generally expected except under quite restricted conditions and unrealistically
simple models of the visual system. I review some of the theoretical
possibilities in relation to available experimental evidence. Relatively simple
mechanistic models provide opportunity for deviations from proportionality,
especially if noise can enter into the neural representation at multiple
stages. The extension to suprathreshold appearance is still more precarious;
yet remarkably, under some experimental conditions, proportionality with
threshold sensitivities holds, in the sense that equal multiples of threshold
match. Observers can voluntarily shift their psychometric functions without losing
sensitivity
Michael Morgan, City University London, Max Planck
Institute for Neurological Research, Cologne Psychometric sensory discrimination functions are usually modeled by
cumulative Gaussian functions with just two parameters, their central tendency
and their slope. These correspond to Fechner’s “constant” and “variable”
errors, respectively. Fechner pointed out that even the constant error could
vary over space and time and could masquerade as variable error. We wondered
whether observers could deliberately introduce a constant error into their
performance without loss of precision. In three-dot vernier and bisection tasks
with the method of single stimuli, observers were instructed to favour one of
the two responses when unsure of their answer. The slope of the resulting
psychometric function was not significantly changed, despite a significant
change in central tendency. Similar results were obtained when altered feedback
was used to induce bias. We inferred that observers can adopt artificial
response criteria without any significant increase in criterion fluctuation.
These findings have implications for some studies that have measured perceptual
“illusions” by shifts in the psychometric functions of sophisticated observers. Sensory, perceptual and response biases: the criterion concept in
perception
Mark Georgeson, Aston University
Signal detection theory (SDT) established in psychophysics a crucial
distinction between sensitivity (or discriminability, d’) and bias (or
criterion) in the analysis of performance in sensory judgement tasks. SDT
itself is agnostic about the origins of the criterion, but there seems to be a
broad consensus favouring ‘response bias’ or ‘decision bias’. And yet,
perceptual biases exist and are readily induced. The motion aftereffect is
undoubtedly perceptual - compelling motion is seen on a stationary pattern –
but its signature in psychophysical data is a shift in the psychometric
function, indistinguishable from ‘response bias’. How might we tell the
difference? I shall discuss these issues in relation to some recent experiments
and modelling of adaptation to blur (Elliott, Georgeson & Webster, 2011). A
solution might lie in dropping any hard distinction between perceptual shifts
and decision biases. Perceptual mechanisms make low-level decisions. Sensory,
perceptual and response criteria might be represented neurally in similar ways
at different levels of the visual hierarchy, by biasing signals that are set by
the task and by the history of stimuli and responses (Treisman & Williams,
1984). The degree of spatial localization over which the bias occurs might
reflect its level in the visual hierarchy. Thus, given enough data, the dilemma
(are aftereffects perceptual or due to response bias?) might be resolved in
favour of such a multi-level model.
|