GRE Psychology Test

Sensation is the detection of physical energy by receptors and its conversion into neural signals; perception is the organization and interpretation of those signals. This chapter covers how sensitivity is measured, how attention selects input, and how objects, depth, and motion are constructed from the retinal image.

Psychophysics, founded by Gustav Fechner, relates the physical magnitude of a stimulus to the experience it produces. The absolute threshold (absolute limen) is the lowest intensity detected on 50 percent of trials. The definition is statistical because detection rises gradually with intensity along an S-shaped psychometric function. Stimuli below the threshold are subliminal; those above it are supraliminal. The difference threshold, or just noticeable difference (JND), is the smallest change detected on half the trials.

The constant k is the Weber fraction. For lifted weights it is about 0.02, so a 100-g standard needs a 2-g change and a 500-g standard needs 10 g. The law fails at extreme intensities. Fechner derived his law by treating each JND as an equal step in sensation; S. S. Stevens replaced that indirect scaling with magnitude estimation, in which observers assign numbers to perceived intensity.

Classical methods treat every “yes” as a pure report of sensation. Signal detection theory (Green and Swets) treats detection as a decision made against background noise. Internal activity varies from trial to trial whether or not a signal is present, so the noise and signal-plus-noise distributions overlap. The observer sets a criterion and says “yes” whenever the internal response exceeds it. Sensitivity and response bias can then be measured separately.

The hit rate is the proportion of signal-present trials on which the observer responds “yes”; the false-alarm rate is the proportion of signal-absent trials on which the observer responds “yes.” Each rate is computed within its own trial type, not over all trials.

Sensitivity is indexed by d′, the distance between the two means in standard-deviation units, computed as z(hit rate) − z(false-alarm rate). Bias is indexed by the criterion location (c) or by β, the likelihood ratio at the criterion. A liberal criterion yields more hits and more false alarms; a conservative criterion yields fewer of both. Payoffs, signal probability, and expectation move the criterion without changing d′.

A receiver operating characteristic (ROC) curve plots hit rate against false-alarm rate as the criterion moves from conservative (lower left) to liberal (upper right). Every point on one curve has the same d′. The diagonal is chance (d′ = 0; area under the curve 0.5); curves bowed toward the upper left show greater sensitivity, up to an area of 1.0. The area is a criterion-free index of sensitivity used in medical diagnosis and memory research. Unequal-variance models often fit ROC data better than equal-variance models.

Sensory adaptation is the decline in responsiveness under constant stimulation. In dark adaptation, sensitivity rises as photopigments regenerate: cones finish within about 10 minutes and rods continue for roughly 30. Light adaptation runs the other way and is faster. Thresholds also depend on the observer: attended stimuli are detected at lower intensities, and motivation or expectation shifts the criterion.

Selective attention favors one input at the expense of others. In Cherry’s dichotic listening task, participants shadow (repeat aloud) the message in one ear. They later report physical features of the unattended message, such as a change of voice, but rarely its meaning or language. The cocktail party effect is the exception: Moray found that about one third of participants noticed their own name in the unattended ear.

Evidence against a strict early filter includes Gray and Wedderburn’s “Dear Aunt Jane” study, in which listeners grouped words by meaning across ears. Divided attention is studied with dual-task designs: performance drops when two tasks draw on the same limited resources (Kahneman’s capacity model) and recovers as one becomes automatic with practice (Schneider and Shiffrin’s automatic versus controlled processing).

Two cortical networks support attention. The dorsal attention network (intraparietal sulcus, frontal eye fields) directs voluntary, goal-driven attention; the ventral attention network (temporoparietal junction, ventral frontal cortex) reorients attention to salient, unexpected events. Right parietal damage can produce spatial neglect: the patient ignores the left side of space, for example eating from only the right half of a plate.

Bottom-up processing builds a percept from sensory features; top-down processing applies knowledge, context, and expectation (a perceptual set). The Gestalt psychologists Wertheimer, Köhler, and Koffka held that the perceived whole differs from the sum of its parts and that the visual system settles on the simplest stable organization, the law of Prägnanz.

Perceptual constancy keeps perceived size, shape, lightness, and color stable while the retinal image changes. Size constancy depends on size–distance scaling, and misapplied distance cues produce the Müller-Lyer, Ponzo, and moon illusions and the Ames room. James Gibson’s ecological approach rejects inference: information in the optic array (texture gradients, optic flow) specifies layout directly, and objects offer affordances for action.

On Gibson and Walk’s visual cliff, most infants old enough to crawl refused to cross the glass over the apparent drop. Campos later found that precrawling infants lowered over the deep side show heart-rate deceleration (interest), whereas fear appears with crawling experience. In motion perception, the phi phenomenon (Wertheimer) is apparent movement produced by stationary lights flashed in succession.