Attention Wolfe et al Ch 7, Werner & Chalupa Ch 75, 78.

Attention

Wolfe et al Ch 7, Werner & Chalupa Ch 75, 78

Attention: Idea of Resource Limitations.

In the example it feels like a limitation. Is this right way to think about it?

Attention is not “brain juice”

– Resource limitations not really a problem or “limitation”, but an inevitable aspectof coordinated goal directed behavior. Ie it wouldn’t help if we were able to process more information – however, learning allows more compact codes.

Potential “bottleneck” role of basal ganglia in action selection:

Nice Chapter by Allport – Foundations of Cog Sci 1989

Attention: a hypothetical internal variable Central idea: selection

This can operate at many levels eg:

- Goal selection: I want a sandwich

- Sub-goal selection: find the peanut butter

- Selection of visual properties associated with PB

- Execution of saccade

- Selection of information at current location of fovea eg location, size etc for programming grasp

Spatial attention, object based attention, auditory attention, etc etc

Question: How does this selection occur?

Given the multiplicity of levels at which attention operates, there is a multiplicity of answers.

Spatial Neglect: lesions of parietal lobe, the frontal lobe, anterior cingulate cortex-profound inability to attend to certain spatial regions

Subcortical level - lesions of the basal ganglia or of the pulvinarthalamic nucleus, which is heavily connected with the parietal cortex

Not sensory or motor: failure to select – therefore thought of as attentional deficit.

Neglect may be object centered (above), eye centered, gaze centered, or body centeredAffects imagined images.

Extinction: image on good side suppresses image on bad side

Note other disorders of attention: schizophrenia (disordered eye movements), ADD

Fronto-parietal network: FEF (frontal eye fields), SEF (supplementary eye fields, and SPL (superior parietal lobule) Note similarity of areas involved in eye movements and attention. Note also, not just spatial attention but attention to objects and features.

From Squires et alFundamentalNeuroscience

Fronto-parietal attentional control system (LIP/FEF)

Cells in LIP do not respond to steadystimuli

Cells respond to behaviorally relevant stimuli

LIP cell responses modulated by reward towards

away

LIP cell responds when relevant cue isin receptive field and when left handis used.Ie modulated by task and hand

Different reward probabilities

Summary: multiple influences on goal (attentional) selection in LIP

LGN receives input from multiple sources including striate cortex, the thalamic reticular nucleus (TRN), and the brain stem. (plus retina)The LGN therefore represents the first stage in the visual pathway at which cortical top-down feedback signals could affect information processing.

fMRI expts show attentional modulation of LGN (even stronger than attentional effects in early visual areas.

Visual SearchHow is attention directed to locations in scenes?

Process by which a region of the peripheral retina is selected as the targetfor a saccade. Called “overt attention” cf “covert attention”

How do we locate objects in visual search?

Easy searches unaffected by increasing numbers of distractors.

Typical experimental paradigm:Easy search – use features eg color to locate the object.

More difficult searches affected by increasing numbers of distractors.

Pop-out search “in parallel” across the visual field.

Versus item-by-item or “serial” search

What is the nature of the limitation in serial search??? Perhaps it is a signaldetection problem.

Think of “pop-out” or “pre-attentive” effects in search as “easy” or high signal/noise ratio.

“Serial” or “attentive” search as low signal/noise. Harder to create a suitable filter.

Posner Paradigm

XX X

P=0.8 P=0.2

Detection performance improved at the cued location, as if attention had shiftedto that location prior to stimulus presentation. (shorter RT or lower threshold)

Cue can be endogenous (arrow) or exogenous (flash at location of target)

Bayesian Approach:

Cueing effect in Posner paradigm without enhanced processing at attended location.Eckstein et al, 2002.

Greater prior liklihood of stimulus at cues location leads tobetter performance (detectability).

Classification Image Technique:

Subject detects a signal in noise

Sort out the False Alarm trials

Add all the images that resulted in false alarms

Reveals the information that led to a false alarm.(Cf reverse correlation technique)

Search Templates

Classification images from subjects in Posner experiment.

The shape of the filter is the same – simply weighted by location probability. Supports Bayesian interpretation of performance, not use of different filters in cued and uncued locations.

Simulation results using classification images

Conclusion:

Think of “pop-out” or “pre-attentive” effects in search as “easy” or high signal/noise ratio.

“Serial” or “attentive” search as low signal/noise. Harder to create a suitable filter.

Idea of attention as biased competition (Reynolds)

What is attention?-Capacity to select information from the environment and select actions to perform

Substantial overlap between circuitry for eye movements and circuitry for spatial attention.

Parietal – frontal network influences visual cortical areas including V1.

LGN may gate incoming visual signals.

Attention appears to act in a way that biases competition between stimuliwithin a receptive field.

Attention is limited - why?Limitations may derive from multiple levels of processing in the brain eg sensory, motor, and sub-cortical circuitry such as basal ganglia.

A more natural situation. How is search guided here?