Attention and Perception I: Forms and Models of Attention; the Gestalt and Physiological Approaches to Perception; and Perceptual Organization — Figure and Ground and the Laws of Organization

Unit 5 opens with "Attention: Forms of attention, Models of attention", then turns to "Perception: Approaches to the Study of Perception: Gestalt and physiological approaches. Perceptual Organization: Gestalt, Figure and Ground, Law of Organization." This chapter sets out the forms of attention and the determinants that capture it, compares the early-selection, attenuation, late-selection and capacity models that grew out of dichotic listening, distinguishes automatic from controlled processing, and then sets the Gestalt account of perception against the physiological one — feature detectors, lateral inhibition and the two visual streams — before listing the laws by which the visual field is organised.

1. Forms and determinants of attention

Forms of attention
FormWhat it is
Selective (focused) attentionAttending to one source while ignoring others — the cocktail-party problem.
Divided attentionAttending to two or more tasks at once; easier when tasks are practised, dissimilar and draw on different resources.
Sustained attention (vigilance)Maintaining attention over long periods to detect rare signals; performance declines within about half an hour — the vigilance decrement first shown with Mackworth's clock test on radar operators.
Alternating (shifting) attentionSwitching between tasks, with a measurable switch cost.
Voluntary versus involuntaryEndogenous, goal-driven (top-down) attention versus exogenous, stimulus-driven (bottom-up) capture, such as the orienting reflex to a sudden noise.
Overt versus covertMoving the eyes to a location versus shifting attention without moving the eyes (Posner's cueing paradigm).

Determinants of attention are external — intensity, size, contrast, movement, novelty, repetition and change — and internal — motives, interests, emotional state and set (a readiness to perceive something). Schneider and Shiffrin (1977) distinguished controlled processing, which is slow, serial, effortful and flexible, from automatic processing, which is fast, parallel, effortless and hard to stop once learned. The Stroop effect — naming the ink colour of the word RED printed in green is slow — shows automatic reading interfering with a controlled task.

2. Models of attention

The models grew from dichotic listening: a different message is played to each ear and the listener shadows (repeats aloud) one of them. Colin Cherry (1953) found that listeners noticed almost nothing of the unattended message — not a switch from English to German — though they did notice a change of voice from male to female. Moray (1959) found that one's own name in the unattended ear was often noticed. The question the models answer is where the bottleneck lies.

The bottleneck debate
ModelClaimEvidence and problem
Broadbent's filter model (1958)Early selection: a sensory buffer holds all inputs briefly; a selective filter passes one channel on the basis of physical features (ear, pitch) to a limited-capacity processor; the rest is lost.Explains Cherry's results but not the own-name effect or Gray and Wedderburn's "Dear Aunt Jane" results, where meaning guided selection.
Treisman's attenuation model (1964)The unattended channel is attenuated (turned down), not blocked; words in a mental "dictionary" have thresholds, and important ones (one's name, danger words) have permanently low thresholds, so they break through.Explains breakthrough by meaning while keeping an early filter.
Deutsch and Deutsch (1963); Norman (1968)Late selection: all inputs are analysed for meaning, and selection happens only at the stage of response or memory.Explains semantic breakthrough but seems uneconomical; people remember very little of unattended messages.
Kahneman's capacity model (1973)Attention is a limited pool of mental effort allocated by an allocation policy influenced by arousal, enduring dispositions and momentary intentions; tasks interfere when their combined demand exceeds capacity.Explains divided attention; Wickens's multiple-resource theory adds that tasks using different modalities or codes interfere less.

Visual attention is often modelled spatially: Posner's spotlight that illuminates a region, and Eriksen's zoom lens, which can widen (with lower resolution) or narrow. Lavie's perceptual load theory reconciles early and late selection: when the attended task is perceptually demanding, distractors are filtered early; when it is easy, spare capacity spills over and distractors are processed.

⚠️ Filter or attenuator?
Broadbent's filter is all-or-none and physical; Treisman's attenuator turns the unattended message down but lets important meanings through. If a stem mentions your own name being heard in the unattended ear, Broadbent's original model cannot explain it.

3. Approaches to perception: Gestalt and physiological

The Gestalt approach (Wertheimer, Köhler, Koffka) holds that perception is of organised wholes, which have properties the parts lack — "the whole is different from the sum of its parts". Wertheimer's phi phenomenon (1912), apparent movement between two lights flashed in succession, was the founding demonstration: the movement is in the perception, not in either stimulus. Organisation is seen as spontaneous and largely innate, and Köhler proposed isomorphism — that the perceived pattern corresponds to a similar pattern of electrical fields in the brain, a hypothesis later experiments did not support.

The physiological approach explains perception by the neural machinery. Lettvin and colleagues found "bug detectors" in the frog's retina (1959); Hubel and Wiesel found simple, complex and hypercomplex cells in the cat's visual cortex tuned to lines of particular orientation (Nobel Prize, 1981). Lateral inhibition among retinal cells sharpens edges and produces Mach bands. Ungerleider and Mishkin (1982) described two visual streams: a ventral "what" pathway to the temporal lobe for object recognition and a dorsal "where" pathway to the parietal lobe for location — which Goodale and Milner reinterpreted as a "how" (vision-for-action) pathway.

4. Perceptual organization: figure and ground, and the laws of organization

The most basic organisation is the segregation of a figure from its ground, described by Edgar Rubin (1915). The figure appears thing-like, in front, with a definite shape and a contour that belongs to it; the ground appears shapeless and continues behind. In reversible figures such as Rubin's vase (a vase or two faces) the two alternate, and only one is seen as figure at a time. Smaller, enclosed, symmetrical and lower regions tend to be seen as figure.

The Gestalt laws of organization
LawElements are grouped when they…
Proximityare close together
Similarityshare colour, shape, size or orientation
Closureform a nearly complete figure — gaps are filled in
Good continuationlie along a smooth line or curve
Common fatemove together in the same direction
Symmetryform symmetrical regions
Common region and connectednesslie within a common boundary, or are physically linked (later additions by Palmer)
Prägnanzcan be seen as the simplest, most regular, stable figure — the overarching law
💡 Everyday examples
A flock of birds rising together is grouped by common fate; the dotted outline of a circle is seen as a circle by closure; a football team in matching shirts is grouped by similarity; and a row of dots spaced in pairs is seen as pairs by proximity.

Key takeaways

  • Attention is selective, divided, sustained (with the vigilance decrement) or alternating, voluntary or captured, overt or covert; automatic processing is fast and parallel, controlled processing slow and serial.
  • Broadbent's filter selects early on physical features; Treisman attenuates rather than blocks; Deutsch and Deutsch select late, after meaning; Kahneman treats attention as limited effort.
  • The Gestalt approach stresses organised wholes and the phi phenomenon; the physiological approach stresses feature detectors, lateral inhibition and the ventral "what" and dorsal "where" streams.
  • Figure–ground segregation (Rubin) is the first step of organisation; reversible figures show that only one region is figure at a time.
  • The laws of organization are proximity, similarity, closure, good continuation, common fate, symmetry and common region, all under the law of Prägnanz.

Practice questions (10)

Attempt each one before opening the answer. Every explanation names the tempting wrong option as well as the right one, because that is where marks are lost.

  1. During dichotic listening, a participant shadowing the right ear notices her own name spoken in the unattended left ear. Which model accounts for this most directly?

    1. Broadbent's filter model
    2. Hubel and Wiesel's feature-detector model
    3. Treisman's attenuation model
    4. The law of Prägnanz
    Show answer

    Answer: C — Treisman's attenuation model

    In Treisman's model the unattended message is turned down, not blocked, and highly important words such as one's own name have permanently low thresholds, so they break through. Broadbent's all-or-none filter predicts the name would be lost.
  2. The decline in detection of rare signals after about half an hour of watching a display is called:

    1. The vigilance decrement
    2. The phi phenomenon
    3. The Stroop effect
    4. Inattentional blindness
    Show answer

    Answer: A — The vigilance decrement

    Mackworth's clock test showed sustained-attention performance falling off within about 30 minutes — the vigilance decrement. The Stroop effect concerns interference from automatic reading, and the phi phenomenon apparent motion.
  3. According to Kahneman's (1973) model, two tasks interfere with each other when:

    1. Their combined demand for mental effort exceeds the available capacity
    2. Both tasks are fully automatic
    3. A filter blocks one channel on physical grounds
    4. They are presented to different ears
    Show answer

    Answer: A — Their combined demand for mental effort exceeds the available capacity

    Kahneman's capacity model treats attention as a limited pool of effort; interference arises when demand outstrips supply. Filtering by physical features is Broadbent's idea, and fully automatic tasks require little capacity and interfere little.
  4. Which of the following are characteristics of AUTOMATIC processing as described by Schneider and Shiffrin? Select all that apply.

    1. Parallel
    2. Fast
    3. Effortful and easily modified
    4. Difficult to suppress once learned
    Show answer

    Answer: A — Parallel; B — Fast; D — Difficult to suppress once learned

    Automatic processing is fast, parallel, effortless and hard to stop once learned (the Stroop effect). Being effortful and flexible describes controlled processing.
  5. Wertheimer's demonstration that two lights flashed in rapid succession are seen as a single light moving is called:

    1. The Müller-Lyer illusion
    2. The phi phenomenon
    3. Mach bands
    4. The autokinetic effect
    Show answer

    Answer: B — The phi phenomenon

    The phi phenomenon (1912) founded Gestalt psychology by showing movement that exists in the perception but in neither stimulus. Mach bands come from lateral inhibition, the autokinetic effect is apparent movement of a stationary light in darkness, and the Müller-Lyer is a length illusion.
  6. Damage to the ventral visual stream (occipital to temporal lobe) would most likely impair:

    1. Reaching accurately for an object
    2. Detecting the location of a sound
    3. Recognising what an object is
    4. Maintaining balance
    Show answer

    Answer: C — Recognising what an object is

    The ventral stream is the "what" pathway for object identification. Guiding reaching and locating objects in space depend on the dorsal "where/how" stream; sound localisation and balance involve other systems.
  7. A flock of birds flying together is perceived as a single group because of the Gestalt law of:

    1. Closure
    2. Figure–ground reversal
    3. Proximity alone
    4. Common fate
    Show answer

    Answer: D — Common fate

    Elements that move together in the same direction are grouped by common fate. Closure fills gaps in incomplete figures, and figure–ground reversal is the alternation seen in Rubin's vase.
  8. Which of the following are properties typically attributed to the FIGURE rather than the ground? Select all that apply.

    1. It appears thing-like, with a definite shape
    2. The shared contour seems to belong to it
    3. It appears to continue behind the other region
    4. It appears to lie in front
    Show answer

    Answer: A — It appears thing-like, with a definite shape; B — The shared contour seems to belong to it; D — It appears to lie in front

    Rubin described the figure as thing-like, in front and owning the contour. Continuing behind, shapeless, is the character of the ground.
  9. Assertion (A): Mach bands — light and dark stripes seen at the border between regions of different brightness — do not exist in the physical stimulus. Reason (R): Lateral inhibition among neighbouring retinal cells exaggerates differences in stimulation at edges. Choose the correct option.

    1. Both A and R are true, and R is the correct explanation of A
    2. Both A and R are true, but R is not the correct explanation of A
    3. A is true, but R is false
    4. A is false, but R is true
    Show answer

    Answer: A — Both A and R are true, and R is the correct explanation of A

    Both are true and R explains A: cells just on the bright side of an edge are less inhibited by their darker neighbours and fire more, and those just on the dark side are more inhibited, creating bands that the light itself does not contain.
  10. Lavie's perceptual load theory predicts that irrelevant distractors will be processed MOST when:

    1. The main task imposes a low perceptual load
    2. The main task imposes a high perceptual load
    3. The distractors are presented to the attended ear only
    4. Attention is fully allocated by an early filter
    Show answer

    Answer: A — The main task imposes a low perceptual load

    Under low load, spare perceptual capacity spills over to distractors, which are then processed (late selection). Under high load, capacity is exhausted by the task and distractors are excluded early.