Transcript Are We Paying Attention Yet?

Are We Paying Attention Yet?
A review of the relation between
attention and saccades
By Travis McKinney
Overview

Corbetta: Covert vs. Overt Orienting -> fMRI
and PET


Moore and Armstrong: FEF stimulation-> V4
response activity
Moore and Armstrong: FEF Stimulation -> V4
discriminability similar to attention effects
Corbetta: What is attention?
Attention: the mental ability to select
stimuli, responses, memories, or thoughts
that are behaviorally relevant among
those that are irrelevant
 How does this relate to foveated vision?

Overt Visual Orienting
Exploring scenes by means of saccadic
eye movements to bring the fovea onto
the stimuli of interest
 Stimuli are processed during fixations
interspersed between saccades

Covert Visual Orienting



Attending to behaviorally relevant stimuli in the
absence of exploratory saccadic eye
movements
The locus of attention is dissociated from eye
fixation
Directing attention toward a location either
voluntarily or reflexively when a stimulus
abruptly appears in the visual field.
Hypotheses for Attention/Eye
Movement Relation
Independence: Attention and eye
movement processes involve entirely
different mechanisms
 Interdependence: Attention and eye
movement processes share resources or
computations at some stage
 Identity: Attention and eye movement
processes involve the same mechanisms

Current View





Attention and eye movements are tightly related
During saccade preparation, oculomotor system
controls location selection even if attention is
directed elsewhere
Direction of attention is dissociable from eye
position during fixations
Findings are do not rule out interdependence or
identity hypotheses
Most findings oppose independence hypothesis
Paradigm




Shifting-attention task: subjects asked to voluntarily shift
attention along a series of locations positioned in left or right
visual field to detect brief visual stimuli with speeded keypress response
Shifting-attention task involves endogenous cueing and
stimuli at attended locations were detected faster than at
unattended locations
Central-detection task: subjects attended to and manually
responded to stimuli in fovea while being presented with the
same series of peripheral stimuli as in the shifting-attention
task
Areas involving covert orienting were localized by subtracting
PET activity recorded during the shifting-attention task from
activity recorded during central-detection task
Shifting Attention Brain Data



Shows differential activity between shifting attention task and
central detection task (should show activity attributed to
attention/covert orienting)
Significant blood flow changes in superior parietal and frontal
cortex
Stronger activation in hemisphere contralateral to attended field
Comparison between studies



Exogenous cueing:
Yellow Foci
Endogenous cueing:
Red Foci
Parietal and Frontal
regions coactivate
when locations are
cued exogenously
and endogenously
Overlap in Activation




There is a very strong overlap in
cortical activation patterns for
spatial cueing and tonic attention
Right hemisphere: activity localizes
along postcentral and intraparietal
sulcus
Left hemisphere: activity straddles
across postcentral and intraparietal
sulcus
Similarity in activation for tonic and
shifting-attention supports idea that
a frontoparietal network is the
source of a selective location
signal, not the site of attentional
modulation
Oculomotor System



In the frontal lobe, activity
centers onto precentral gyrus
A second cluster of activity
appears at posterior tip of
superior frontal sulcus
In the parietal cortex activity
is distributed near
intraparietal and postcentral
culcus and adjacent gyri and
extends towards the
precuneus
Brain Activity: Attention vs.
Saccades


Eye movement activity: evident dorsally in the
right precuneus and left postcentral gyrus
Attention activity: evident ventrally in the
intraparietal sulcus
Brain Activity: Attention vs.
Saccades




All three major sites of activation for attention (intraparietal,
postcentral, and precentral) show convergent activation
during eye movement
Presence of attentional activity in frontal eye fields indicates
that attention related signals can be recorded in an area
strongly implicated in voluntary oculomotor planning
This data supports the interdependence hypothesis and does
not rule out the identity hypothesis
This data does NOT support the independence hypothesis
Attention vs. Saccades in same Subject



Single subject scanned during
covert shifting-attention task and
during an overt shifting task
Left and right visual field were
tested independently
Attentional activation and
saccadic eye movement
activation localized to identical
brain regions for both left and
right visual fields
Monkey Studies: Covert
Orienting




During tasks emphasizing exogenous cueing at a location a
suppressive type of modulation has been found
Neurons in parietal cortex gave a brisk response to a cue when
flashed in visual field
Responses to subsequently presented probe stimuli at attended
locations were either
 Unaffected by cue: 48%
 Depressed by the cue: 42%
 Enhanced by the cue: 10%
This suggests that the sensitivity of parietal neurons decrement at
a given location after that location has been selected
Monkey Studies: Overt
Orienting



Oculomotor signals have been measured in many
areas of the macaque brain (FEF, dorsolateral
prefrontal cortex, caudate and superficial layers of
superior colliculus, etc.)
The neural response to visual stimuli is enhanced
when the stimulus is the target of a saccadic eye
movement
Neurons in area LIP respond to visual stimuli and
show preparatory oculomotor activity, indicating
that attention and eye movement signals are tightly
related at the neuronal level
Monkey Studies: Overt
Orienting





Monkeys trained in a spatially cued oculomotor task
Saccadic reaction times for cued locations were faster than
uncued locations for exogenous and endogenous cueing
Electrical stimulation with microcurrents produced a
displacement of the constant saccade vector in the direction
of the cued location
Normally, this stimulation would generate a saccadic eye
movement of constant direction and amplitude
Therefore attentional shifts independent of eye movements
still lead to modification of evoked saccades
Moore & Armstrong: Nature 2003
Examined interaction between saccade
preparation and visual coding with
microstimulation of frontal eye fields
 Measured effect of microstimulation on
neural activity in extrastriate visual cortex
 Microstimulation was below the level
necessary to evoke a saccadic movement

FEF Microstimulation



FEF is involved in the selection
of visual targets for saccades
Electrical stimulation of FEF
evokes short-latency saccades
in human and non-human
primates
Stimulation below threshold
does not evoke saccades, but
biases the selection of eye
movements and can improve a
monkey’s ability to covertly filter
visual stimuli
FEF Microstimulation




Stimulation applied 200-500
ms after appearance of visual
stimuli
Neuron responds to stimulus
in RF, but adapts within ~250
ms
Microstimulation enhanced
response with respect to
control condition (figure 2a)
Microstimulation had no effect
when no stimulus was present
in RF (figure 2b)
FEF Microstimulation




When a stimulus is presented in
RF, microstimulation enhanced V4
responses (figure 3a)
When no stimulus is present in RF,
microstimulation has no effect on
V4 population response (figure 3a)
The preferred stimulus yields a
greater response enhancement
than the non-preferred stimulus
The largest response
enhancement occurs when the
preferred stimulus is presented in
the RF with a distractor outside the
RF (figure 3b)
Not always enhanced?



V4 response suppression occurred when preferred
stimulus was presented in the RF and a distractor was
presented outside of the RF in cases when evoked
saccade would not have been in direction of RF (figure
3b)
FEF microstimulation appears to have activated a
network that controls gain of visually driven signals
Results show that activation of this network biases eye
movement selection as well as strength of visual cortical
signals, revealing a common network for visual and
oculomotor selection (supporting identity or
interdependence hypotheses)
Armstrong & Moore: PNAS 2007




Voluntary attention improves the discriminability of
visual cortical responses to relevant stimuli
Recent work implicates the frontal eye field in
driving spatial attention
Subthreshold microstimulation enhances V4 neural
response, but it is unknown whether the
enhancements include improved visual-response
discriminability (part of voluntary attention)
Armstrong and Moore explored response
discriminability in this paper
Solid as a ROC?



Used receiver-operating characeristic (ROC)
analysis to quantify how well v4 neurons could
discriminate two stimuli
Several hundred ms after visual stimulus onset,
response adaptation had reduced the
discriminability of V4 neurons to different stimuli
Subthreshold microstimulation of FEF restored
response discriminability
Paradigm



One of two differently oriented bars was
presented in RF of single V4 neurons in
monkey
Monkey is performing a passive fixation task
AROC is area under ROC curve, and is
performance expected of an ideal observer
making a decision about RF stimulus
orientation based on neuron's response
Visual Response Discriminability




V4 response to 45º bar is higher
than response to 135º bar
(figure 1a)
AROC curve shows probability of
perfect observer being able to
discriminate stimuli based on V4
response (figure 1b)
Neurons with stimulus tuning
during onset analysis window
show higher discriminability
(figure 1c)
Discriminability decrease as a
function of time (figure 1d)
V4 Response Discriminability




V4 response is enhanced
with stimulation only when
visual stimuls is oriented at
45º (figure 2a)
AROC is significantly higher
for stimulation than for
control (figure 2b)
Discriminability is higher for
stimulation neurons than
control neurons (figure 2c)
Effect of microstimulation
on discrimination positively
correlated with change in
discriminability between
onset and late trials (figure
2d)
Visual Stimuli Spatial Alignment



Subset of neurons tested with RF stimuli either
spatially aligned or misaligned with saccade
vector possibly evoked from stimulation site
Microstimulation enhanced neuronal response
discriminability for visual stimuli appearing at
aligned RF position
Microstimulation had no effect on neuronal
response discriminability for visual stimuli
appearing at the misaligned RF position
Effect of Spatial Alignment


When RF stimulus is
aligned, stimulation
enhances discriminability
(figure 3a)
When RF stimulus is
misaligned, stimulation
has no effect on
discriminability (figure
3b)
Response Reliability



Regarding the relationship
between response magnitude
(spike count) and variance,
power terms and coefficients
were not significantly different for
stimulation vs. control (figure 4a)
Stimulation enhanced response
to preferred stimuli, but not to
non-preferred stimuli (figure 4b)
FEF microstimulation, like
voluntary attention, improves
response discriminaability without
altering response reliability
Timing and Simulated Phosphenes




In the first poststimulation time bin
(first 40 ms), discriminability had
already been significantly increased
(figure 5a)
Examined influence of simulated
phosphene on response
discriminability (figure 5b)
Simulated phosphene disrupted
response discriminability 70ms after
phosphene onset
Simulated phosphene is temporarily
masking the stable RF stimulus
when presented simultaneously
The End

Thanks
The Human Brain
Precentral sulcus
Central sulcus
Postcentral sulcus