Lec 5 - Attention and Working Memory

Selective attention

The cognitive and neural process of prioritizing certain sensory inputs or features for enhanced processing while ignoring others

• modulates firing rate, response gain, and noise in neurons to influence perception and behavior

At any moment, we process only a tine portion of available sensory input

Spatial attention

Type of selective attention directed to a location in space

Feature-based attention

Type of selective attention directed to a non-spatial feature (eg. Color or orientation)

Object-based attention

Type of selective attention directed to an object or object category

Covert Attention

Type of selective attention directed to a feature without moving the eyes

Endogenous (voluntary) attention

Type of selective attention controlled by intention (top-down) and goals

Exogenous (voluntary) attention

Type of selective attention drawn to some location, feature, or object

• stimulus drive (bottom-up)

Topdown/bottom-up

Goal directed/stimulus driven dimension of selective attention

Divided/focused

dimension of selective attention where mental resources are distributed across multiple tasks/concentrated on a single stimulus while ignoring all others

Automatic/controlled

dimension of selective attention That requires little conscious control/requires conscious effort (but more flexible for novel or complex tasks)

Non-conscious/conscious

dimension of selective attention Of subliminal processing/intentionally focusing on specific info

Peripheral/automatic

dimension of selective attention that is automatic, stimulus driven, reflexive attention to events on the outside of sensory space/voluntary, goal-directed, controlled attention focused on specific locations, objects, or features that requires cognitive effort

Preattentive/attentive

dimension of selective attention that is nonconscious and fast/ slower and conscious

Discrete/continuous

dimension of selective attention where processing switches from unfocused to highly selective at distinct point in time/attentional selectivity gradually improves over time

Serial/parallel

dimension of selective attention That processes one stimulus at a time/processes multiple stimuli simultaneously

Local/global

dimension of selective attention that focuses on smaller, discrete elements of a scene/focuses on overall structure or context formed by the elements

Perceptual/cognitive

dimension of selective attention filtering and processing relevant to physical demands of a task/selecting and processing info at higher conceptual level

Signal quality

A division of selective attention studies focused on neuronal firing rates, contrast response functions, etc.

Representation

A division of selective attention studies focused on, for example, how many neurons represent one stimulus dimension or another

Response gain

The increase in a neuron’s firing rates when attention is directed toward its preferred stimulus, effectively amplifying its response without changing its selectivity

Spontaneous Firing Rate

The baseline rate at which a neuron fires action potentials in the absence of any deliberate external stimulus

Background noise

Attention can increase signal quality by decreasing what?

Contrast gain mechanism

A way attention modulates neuron responses by increasing a neuron’s sensitivity to differences in stimulus contrast, effectively making weak signals more detectable without changing the maximum firing rate

Response gain mechanism

A way attention modulates neural responses by multiplicatively increasing the firing rate of neurons across all stimulus strengths, amplifying responses to preferred stimuli without changing sensitivity to contrast

Activity gain mechanism

A form of attentional modulation where attention increases a neuron’s overall baseline firing rate (spontaneous activity), shifting the entire response curve upward regardless of stimulus strength.

Feedback circuitry

Neural pathways from higher-order areas back to earlier sensory areas. They modulate synaptic weights and neuronal responses, integrating context, expectations, and attention. This top-down influence refines sensory processing and shapes perception beyond raw input.

Corticothalamic loops

Reciprocal circuits between cortex and thalamus. Feedforward pathways relay sensory input from thalamus to cortex, while cortical feedback regulates these feedforward processes by modulating synaptic weights, gain, and filtering. This dynamic loop shapes attention, sensory precision, and flexible information processing.

Attentional channels

Neural pathways that enhance relevant inputs and suppress irrelevant ones, prioritizing information by modulating synaptic weights and gain. They often operate through oscillations of local field potentials (e.g., alpha, gamma rhythms), which coordinate timing across brain areas to regulate feedforward processing and improve selective attention.

Matched filter theory

The idea that neurons (or neural populations) are tuned to detect specific, behaviorally relevant patterns of input. Each neuron maximizes sensitivity to particular stimulus features (e.g., orientation, frequency). This tuning increases efficiency, reduces noise, and allows the brain to selectively amplify important signals.

• warps retinotopic maps as receptive field is shifted to target of attention (V4 shifts ~10%-20%)

  • Consistent with magnitude of attentional modulation on V4 firing rate (~15%)

• BOLD responses also move toward attended region

Default mode network

Network of brain regions most active when person is engaged in internal thoughts (rumination, daydreaming, reminiscing, etc)

• decreases activity during attention-demanding, goal-directed tasks

• Voxels here tend to shift away from attended category

Unattended categories

Attended categories are more easily decoded from brain activity than what?

Working memory

A limited-capacity system for temporarily holding and manipulating information needed for ongoing tasks. It involves prefrontal and parietal networks, supports reasoning, decision-making, and goal-directed behavior, and is maintained through persistent neural activity and recurrent circuits

• may be a form of attention

  • Both rely on sustained activity during delays in their experiments

Changing weights

An attentional gain change is analogous to this in a neural network