Attention: How the Mind Selects, Binds, and Shifts Information

You search a crowded station for a friend in a red coat. A red advertisement catches your eye, your phone vibrates, and someone nearby says your name. Your senses receive more information than you can deliberately examine. Attention helps determine what receives priority, what gets combined into an object, and when you move on.

These notes combine Lecture 7 on normal attention with Lecture 8 on disorders of spatial awareness. Clinical cases help separate abilities that usually work together; they are not self-diagnosis exercises.

Learning goals and Feynman practice

Explain selection in listening, distinguish controlled processing from stimulus-driven capture, interpret visual-search and cueing results, and compare blindsight with neglect. Use these mechanisms to explain a new example rather than merely remember their names.

Read a section, close it, and teach the idea to a beginner. State the observation, explain the proposed mechanism, and name a limitation. When your explanation breaks, return to that section and repair it. Try each question before opening its answer.

Part 1: Mechanisms and limits of normal attention

1. Attention is selective priority

William James’s 1890 account emphasized selecting one object or thought while withdrawing from alternatives. In modern terms, attention prioritizes processing: some locations, objects, features, or moments receive greater weight than others.

The gatekeeper metaphor is useful but incomplete. Attention often improves conscious access and memory encoding, yet some unattended information still affects processing. Attending to something also does not guarantee that it becomes a durable memory. Selection, awareness, and later recall require different measurements.

Think of a limited editorial team: many stories arrive, but only some receive detailed investigation. The analogy illustrates priority and competition; there is no little editor inside the brain directing every operation.

Dichotic listening: what survives outside the focus?

In Cherry’s classic shadowing task, different messages arrive at the two ears, and the listener repeats one message aloud. Listeners generally report little of the unattended message’s meaning while sometimes detecting conspicuous acoustic changes, such as a voice becoming a tone.

Poor reporting does not establish that every unattended word receives zero semantic processing. Results depend on task demands, stimulus relevance, and occasional shifts of attention. Detecting a personally relevant item could reflect processing outside the main focus, a brief attentional shift, or both; a design must distinguish these possibilities.

Teach it back: Why does remembering the unattended speaker’s voice but not the message suggest selective processing rather than a disconnected ear?

2. Controlled versus automatic is not the same distinction as top-down versus bottom-up

Distinction	What it concerns	Example
Controlled versus automatic	How dependent processing is on deliberate control, practice, and available capacity.	A beginner decodes a word deliberately; a skilled reader often reads it rapidly.
Top-down versus bottom-up	Whether priority is influenced by goals or by incoming stimulus properties.	Looking for red coats versus orienting toward an abrupt flash.

These dimensions interact but should not be collapsed. A practiced, goal-directed action can be relatively automatic. A salient stimulus can capture attention without being wholly unconscious. Automaticity is usually a matter of degree and conditions, not a guarantee of zero effort or zero interference.

Stroop interference

Naming the ink colour of the word BLUE printed in red is usually slower or more error-prone than naming the colour in a suitable comparison condition. A well-practiced reading response competes with the instructed colour response.

A useful analogy is two people giving different answers at once, with one voice much more practiced. The mechanism concerns competing task-relevant processing, not two literal speakers in the head.

Research on highly hypnotizable participants found reduced Stroop interference after suggestions to treat words as meaningless symbols. The supplied outline’s 157-to-5 ms contrast should be treated as a study-specific figure, not a universal effect or personal prediction. Without the original lecture bibliography, these notes do not use that number as an established benchmark. Related experiments show that suggestion can also matter without formal hypnotic induction. [1, 2]

Teach-back check: why does a smaller Stroop effect not prove that reading has been permanently switched off?

The effect is a difference between task conditions under particular instructions. Strategies, expectations, response competition, and processing changes may contribute. A temporary experimental difference does not establish a permanent loss of reading or a universal effect across participants.

3. Visual search: finding a feature versus a combination

Imagine finding a red circle among blue circles. Now find a red X among red Os and black Xs. In the second display, neither redness nor X-shape alone identifies the target.

Search condition	Typical observation	Interpretation and limit
Distinctive feature	A target may pop out, with little increase in search time as distractors increase.	Compatible with efficient selection; not every single-feature search is equally easy.
Feature conjunction	Search time often increases more with display size.	Selecting the correct feature combination can require more processing; it need not be a rigid item-by-item scan.

Set size is the number of items in a display. A search slope describes the change in response time per additional item. For a made-up example, increasing set size from 8 to 16 and response time from 600 to 760 ms gives a slope of 20 ms per item. These are teaching values, not a published result.

A flat slope does not mean processing takes no time. A steep slope alone does not prove serial processing: similarity, crowding, grouping, target probability, and guidance can affect search. Many conjunction searches become efficient when a feature strongly narrows the candidates.

Feature binding and illusory conjunctions

Treisman and Gelade’s feature-integration theory proposed that attention helps combine separately represented features into the appropriate objects. [3] With a brief red X and green O display, especially under competing demands, someone may report a red O: an illusory conjunction.

“Attentional glue” is a metaphor for assigning features to the right object or location. It does not mean unbound colours physically float through the brain, or that every binding operation requires deliberate consciousness.

Teach it back: Why is a red X difficult to find when some distractors are red and others are Xs? Give a situation where colour could still guide the search.

4. Bottlenecks in time and space

The attentional blink is reduced report of a second target that appears shortly after a first in a rapid stream, often around 200–500 ms later. It concerns temporal selection and access under the tested conditions, rather than an actual eye blink or a shutdown of all visual processing. At very short separations, the second target can sometimes be spared. [4]

Multiple object tracking tests spatial and dynamic limits. Several identical objects are briefly marked as targets, move among distractors, and must later be identified. People often track only a few reliably, but capacity varies with speed, spacing, crowding, motion patterns, and task demands. There is no universal fixed object count for every situation.

Teach-back check: why might the same person track four slow, well-spaced objects but fail with three fast, crowded ones?

Tracking depends on maintaining target identity through motion and competition. Speed and proximity can increase confusion and processing demands. A capacity estimate describes performance under specified conditions, not a permanent number of mental slots proven by one task.

Part 2: Disorders of attention and spatial awareness

5. Blindsight: performance and reported seeing can separate

Following damage to primary visual cortex, V1, some patients retain above-chance localization or motion discrimination in an affected visual-field region despite reporting no normal visual experience there. Others describe vague residual sensations. A scotoma is a region of impaired vision; V1 injury may also produce a larger field loss such as hemianopia.

The main geniculostriate route involves the LGN and V1. Residual performance can involve pathways bypassing damaged V1, but not exclusively the superior-colliculus–pulvinar route. Studies also implicate direct LGN–extrastriate connections. Mechanisms vary with injury, spared tissue, task, and reorganization. [5]

Researchers compare forced-choice performance with awareness reports while controlling fixation and stimulus placement. Above-chance guessing does not establish ordinary conscious sight or mean every patient can navigate obstacles accurately.

Teach it back: What two measurements are needed to demonstrate a dissociation between visual performance and reported experience?

6. Neglect: available information does not receive normal priority

Hemispatial neglect is impaired attention, response, or representation for one side of space or objects that cannot be explained solely by a primary sensory or motor deficit. Left neglect commonly follows right-hemisphere stroke involving parietal and connected frontal, temporal, subcortical, or white-matter networks. It is not confined to one cortical point.

Examples include leaving food on the left of a plate or omitting parts of a drawing. Neglect can affect visual, tactile, auditory, and exploratory behaviour, and sometimes remembered scenes. Different forms need not occur equally in one person.

Anosognosia, the standard spelling, means impaired awareness of a deficit, such as paralysis. It can accompany neglect but is distinct from it. Eating only from one side of a plate can indicate neglect; denying paralysis addresses awareness of motor impairment. Neither behaviour alone establishes the other condition.

Assessment	Possible observation	Interpretive caution
Line bisection	Marking the midpoint too far toward the attended side.	Bias depends on line position and task conditions.
Cancellation	Omitting targets on one side.	Vision, motor ability, and search strategy also matter.
Figure copying	Leaving out one side of a scene or object.	Compare perceptual, constructional, and attentional contributions.

A single normal test does not exclude neglect. Assessments sample different demands and should be interpreted with everyday behaviour and clinical findings.

Teach-back check: why is neglect not simply “the left eye cannot see”?

Both eyes normally provide information about both sides of visual space. Neglect can affect multiple senses, object-relative coordinates, and imagined scenes. It concerns attentional or representational bias rather than loss of input from one eye, although sensory field defects can coexist.

7. Posner’s cueing task: disengage, move, engage

In a spatial cueing experiment, the participant maintains central fixation. A cue indicates one location, then a target appears. A valid cue indicates the eventual target location; an invalid cue points elsewhere. Faster responses after valid cues can reveal covert attention, meaning a shift of priority without an eye movement.

A classic account separates disengaging from the current focus, moving attention, and engaging at a new location. After right parietal injury, a particularly large delay can occur when a cue draws attention rightward and the target then appears leftward. Posner and colleagues interpreted this as evidence for impaired disengagement in orienting toward a contralesional target. [6]

The key contrast is the extra cost of an invalid cue, not merely slow responding on the left. Sensory, motor, alertness, and eye-movement differences need consideration. This influential mechanism does not mean every patient has perfectly normal moving and engaging operations or that disengagement explains all neglect.

Teach it back: Why compare a left target after right and left cues rather than compare left and right targets only?

Self-check: separate a baseline disadvantage from a cueing cost

Comparing different cue conditions for the same target side helps estimate the additional cost of redirecting attention. A general left-side detection or response problem could otherwise look like a disengagement problem. Further controls are still needed to interpret the effect.

8. Object-based coordinates and residual processing

A rotating object can carry an attentional bias

In Tipper and Behrmann’s rotating-barbell experiments, targets appeared at connected ends of an object. Rotation changed which end occupied each side of external space. Detection costs could follow the object’s previously disadvantaged end, affecting performance even after it moved rightward. [7]

This supports an object-based contribution alongside spatial coordinates. It does not prove that all neglect rotates with every object or operates only in object-centred coordinates. The comparison with unconnected objects and control of eye movements helps distinguish object history from current screen location.

Imagine putting a sticker on one end of a rotating baton. Following that end as it moves differs from always selecting whatever is currently on the left. Object-based attention can preserve information about the end across movement.

The house-on-fire case

Marshall and Halligan described a patient who judged two house drawings identical despite flames drawn on the neglected side of one, yet preferred the non-burning house as a place to live. [8] The case suggests that neglected information can influence a choice without supporting an explicit report of the difference.

It is a single-case dissociation, not proof that all unattended material receives fully intact semantic analysis. Repeated trials, stimulus controls, task wording, and measures of awareness matter. A preference shows a behavioural influence; it does not reveal the full content of an unreported experience.

Teach it back: Why does choosing the safer-looking house establish less than consciously identifying where the flames are?

9. Worked comparison: similar omissions, different mechanisms

Observation	Hypothesis to investigate	Evidence needed
Poor report of a rapid second target.	Temporal attentional limitation.	Accuracy across target separations and matched task conditions.
Above-chance discrimination in a field with no normal reported sight.	Blindsight or residual vision.	Objective performance, awareness reports, fixation, and lesion assessment.
Omissions on the left across drawing and touch tasks.	Spatial neglect.	Multimodal testing and sensory and motor controls.
Denial of an established paralysis.	Anosognosia for motor impairment.	Assessment of insight and relevant alternative explanations.

The table is a way to organize evidence, not a shortcut to diagnosis. One behaviour can have several causes, and different conditions can coexist.

Retrieval workshop and spaced review

From memory, explain dichotic listening, Stroop interference, search slopes, the attentional blink, and invalid-cue costs. For each, state what the participant does and what measurement changes. Then explain why blindsight, neglect, and anosognosia are distinct.

Repair: “Automatic means bottom-up, unconscious, and effortless”

Automaticity concerns reduced dependence on deliberate control and capacity under particular conditions. Goal-directed routines can become automatic, and stimulus-driven capture is a different dimension. Awareness and effort must be assessed rather than assumed.

Repair: “A flat search slope proves the whole display is consciously perceived”

A flat slope indicates little added search time as set size increases in that task. Efficient target selection does not establish detailed conscious access to every distractor or a particular processing architecture by itself.

Apply: design a test of object-based versus screen-based neglect

Move a connected object’s initially left and right ends to opposite screen locations, then compare detection at each end with static and unconnected controls. Track fixation and compare the effects of current location with object history. A cost that follows an end supports an object-based contribution rather than a fixed screen-coordinate account alone.

Review tomorrow and in a week. Score each explanation: 0 for a remembered label, 1 for a definition, 2 for a mechanism plus example and limitation. Repair the weakest explanation first.

One-minute explanation: Attention prioritizes some processing over alternatives. Practice, goals, stimulus properties, and competition shape that selection. Limits appear in searching, binding, tracking, and rapid target report. Clinical dissociations show that sensory processing, orienting, conscious report, and awareness of impairment can separate. Good explanations identify the task and comparison before assigning a mechanism.

Sources and further reading

The supplied numbered citations did not include a bibliography. These primary studies support the experimental and clinical examples; the station, baton, and numerical slope examples are teaching aids.

1. Raz, A., Fan, J., and Posner, M. I. (2005). Hypnotic suggestion reduces conflict in the human brain.
2. Raz, A., et al. (2006). Suggestion reduces the Stroop effect.
3. Treisman, A. M., and Gelade, G. (1980). A feature-integration theory of attention.
4. Raymond, J. E., Shapiro, K. L., and Arnell, K. M. (1992). Temporary suppression of visual processing in an RSVP task: an attentional blink?.
5. Ajina, S., et al. (2015). Human blindsight is mediated by an intact geniculo-extrastriate pathway.
6. Posner, M. I., Walker, J. A., Friedrich, F. J., and Rafal, R. D. (1984). Effects of parietal injury on covert orienting of attention.
7. Tipper, S. P., and Behrmann, M. (1996). Object-centered not scene-based visual neglect.
8. Marshall, J. C., and Halligan, P. W. (1988). Blindsight and insight in visuo-spatial neglect.