Spatial Attention, Attentional Networks, and Divided Attention

Historical context: Extended classic attention research of the 1980s–1990s to spatial attention.
Basic trial sequence
- Central fixation cross → participant keeps eyes locked on it.
- Optional spatial cue (usually an arrow) appears.
- Target stimulus follows at some screen location.
- Task: either (a) state where the target appeared or (b) simply report its presence.
Cue validity manipulations
- Valid cue: correctly predicts location (e.g., $80\%$ of trials).
- Invalid cue: misleads to the wrong location.
- Neutral condition: no directional cue, only fixation cross.
Typical findings
- Valid cue → faster reaction time (RT) & often better accuracy vs. neutral.
- Invalid cue → slower RT vs. neutral (cost of mis-allocation).
- Demonstrates expectation-driven pre-allocation of attentional resources.
Metaphor: Attention behaves like a spotlight that illuminates a chosen spatial region, leaving non-illuminated areas in a processing “darkness.”
Connection to earlier material: Mirrors previously discussed priming effects in non-spatial domains (e.g., semantic or object priming)—now the prime is place.

Saccadic latency
- Typical eye shift occurs at ≈ $180\text{ ms}$ after a spatial cue.
Frontal eye fields (FEF)
- Bilateral frontal-lobe regions controlling ocular muscles.
- Initiate, guide, and inhibit saccades.
Attention vs. eye position
- Spatial attention can shift internally faster than the eyes move.
- Event-related potential (ERP) studies show attentional shifts as early as $150\text{ ms}$ post-cue—preceding physical eye movements.
- Implies that attentional “spotlight” can disengage from current gaze, previewing upcoming saccade targets.
Conceptual link: Attentional shifts may be an internalized form of planned eye movements, sharing circuitry but operating on different time-scales.

Task: Participants attend left vs. right visual field while no actual stimulus is shown (only central fixation).
Findings
- Attending left increases BOLD signal in right occipital lobe (contralateral mapping).
- Attending right shows complementary left-hemisphere pattern.
Significance: Mere intention to attend activates early visual cortex at the relevant retinotopic coordinates—underscoring top-down modulation.

Anatomical split from V1/V2
- Dorsal stream (“Where” pathway)
- Projects to parietal lobes.
- Encodes spatial position, motion, depth.
- Ventral stream (“What” pathway)
- Projects through temporal lobe.
- Encodes object identity (faces, houses, tools, etc.).
Integration requirement: Motor planning demands merging object and location information so actions target the right thing in space.

Meta-analysis (textbook figure) aggregates imaging studies.
Dorsal attentional network (DAN)
- Nodes: Posterior parietal cortex + Frontal eye fields.
- Function: Top-down goals—prepare to attend particular locations.
- Aligns with dorsal “Where” visual stream.
Ventral attentional network (VAN)
- Nodes: Temporal lobe areas + ventral lateral prefrontal cortex.
- Function: Bottom-up capture by salient objects or features.
- Aligns with ventral “What” stream.
Hemispheric symmetry: Both networks exist bilaterally; interactions resolve competition between goal-driven and stimulus-driven signals.

Lesion profile: Damage to right parietal lobe → neglect of left visual field.
Clock-drawing example
- Patient draws all clock numbers bunched on the right side—object knowledge intact, spatial layout compromised.
Berman & Trippier rotation task
- Setup: Red circle in right VF, blue circle in left VF, each containing objects.
- Neglect patient initially reports only right-field (red) objects.
- Circles rotate: red (original objects) now in left VF—but patient can still track & name them.
- Interpretation: Once dorsal network locks onto an object, ventral/object-based tracking can maintain attention even when spatial coordinates shift into neglected hemifield.
Implication: Distinct yet interactive systems—object-based attention can partially circumvent spatial neglect.

Definition: Simultaneous performance of two tasks demanding attention to separate input streams.
General outcome: Dual-task performance < single-task performance (RT ↑, accuracy ↓) because of competition for limited cognitive resources.
Resource theory distinctions
- Task-specific resources: unique pools (e.g., verbal working memory) → interference highest when both tasks tap the same pool.
- Task-general resources: shared central capacity (executive control, response selection, etc.) → always at risk of overload.
Classic study: Allport et al., 1972
1. Condition A: Dichotic listening + recall of unattended words from opposite ear → high interference.
2. Condition B: Shadow auditory words + recall visually presented words → less interference (different modality reduces overlap).
3. Condition C: Shadow auditory words + recall pictures seen on screen → minimal interference (distinct representational codes: phonological vs. pictorial).
Practice & automaticity
- Repetition converts controlled processes into automatic ones.
- Automatized tasks demand fewer resources → diminished dual-task cost.
- Supports strategies for skill learning (e.g., driving while conversing becomes easier with experience).

Attention as proactive filtering: Expectations bias early sensory cortex before stimuli appear.
Temporal hierarchy: Neural markers of covert attention precede physical eye movements.
Functional segregation yet integration
- Dorsal (where/top-down) & ventral (what/bottom-up) streams provide anatomical basis for two attentional networks.
- Effective behavior requires their coordination (e.g., grasping a cup at the correct spot).
Clinical insights
- Spatial neglect underscores dissociation between object and location processing.
- Rehabilitation may leverage intact object-based systems to compensate for spatial deficits.
Resource limitations
- Cognitive capacity is finite; performance reflects allocation efficiency.
- Training and automatization free capacity, illustrating neuroplasticity in service of multitasking.