Cognition & Language

Cognition & Language

Part 1 of 4 — Attention & Information Processing

The Information-Processing Frame

Cognition treats the mind as a limited-capacity processor: input → attention filters → working memory operations → long-term storage → output. Every phenomenon in this lesson is about where the bottleneck sits and what slips past it.

Selective Attention — One Channel at a Time

Dichotic listening (different messages to each ear, shadow one): people report almost nothing from the unattended ear — except physical features (voice gender, pitch) and, sometimes, their own name (the cocktail party effect).

Three classic accounts, ordered by where the filter sits:

What Unattended Really Costs

• Inattentional blindness: fully visible events go unseen when attention is engaged elsewhere (the gorilla walking through the basketball passers)
• Change blindness: large changes across a visual disruption go unnoticed — perception keeps far less than it seems to
• Both prove that looking is not seeing: awareness requires attention, not just retinal registration

Divided Attention, Automaticity & the Stroop Effect

• Controlled processing: effortful, serial, capacity-limited (novel or complex tasks)
• Automatic processing: fast, parallel, nearly effortless — built by extensive practice (reading, for literate adults)
• Dual-task performance succeeds when tasks differ in modality/resources or when one is automatic; it collapses when two controlled tasks compete (why phone conversations impair driving even hands-free — the interference is central, not manual)
• Stroop effect: naming the ink color of the word RED printed in blue is slow because automatic word-reading delivers a competing response that must be suppressed. Stroop interference is the standard laboratory index that a process has become automatic and involuntary.
• Task switching: switching attention carries a measurable time cost; "multitasking" is rapid alternation, with a per-switch penalty

Processing Speed Signatures in Data

• Reaction-time (RT) increases that are linear in the number of items suggest serial search; flat RT curves suggest parallel/automatic processing ("pop-out" of a red item among greens)
• Practice shifts a task from the serial to the flat pattern — that shift, plus Stroop-like intrusions, is how passages measure automaticity

Worked Example — An Attention Passage, Fully Read

Passage. Participants shadow a story played to the right ear while a word list plays to the left. Afterward, recognition of left-ear words is at chance — but participants' skin conductance rises when previously fear-conditioned words occur in the left ear during shadowing, and 34% of participants detect their own name there. In Experiment 2, novices and expert typists type while repeating aloud a spoken passage; novices' typing accuracy collapses, experts' barely changes. In Experiment 3, expert typists show large interference when the spoken passage is replaced by a second manual task (sorting cards by suit).

Step 1 — locate the filter. Chance recognition says unattended content did not reach explicit memory. But the skin-conductance response to conditioned words and the name breakthrough show meaning was analyzed at least shallowly for select, low-threshold items — too much analysis for Broadbent's strict early filter, exactly what Treisman's attenuation predicts. Cite attenuation, not late selection: late selection predicts widespread semantic processing, and chance recognition argues against that.

Step 2 — read the expertise dissociation. For experts, typing has become automatic — it no longer consumes the central resources that speech shadowing needs, so the two proceed in parallel. Novice typing is controlled processing, and two controlled tasks overload capacity.

Step 3 — explain Experiment 3's reversal. Automaticity is resource-specific, not absolute: two MANUAL tasks now compete for the same motor/spatial resources, restoring interference even in experts. This is the multiple-resource refinement the MCAT loves — automatic does not mean cost-free under every pairing.

Step 4 — anticipate the application item. "What does this predict about hands-free phone use while driving?" Interference persists, because conversation and hazard monitoring compete for central attention, not for hands — the passage's novice-typist logic transferred to the road.

Attention & Automatic Processing 🎯

Key Takeaways — Part 1

• Filter theories: Broadbent = early strict block; Treisman = early attenuation with low-threshold breakthrough (the cocktail party answer); late selection = filtering after meaning
• Inattentional and change blindness prove awareness needs attention, not just fixation — looking is not seeing
• Controlled processing is serial, effortful, capacity-limited; automaticity comes from practice, runs involuntarily, and is measured by Stroop intrusion and flat RT-by-set-size curves
• Dual-task success requires different resources or an automatic member; hands-free phone impairment is central, not manual
• Data habit: linear RT slope = serial search; flat slope = parallel pop-out; practice converts the first into the second

Cognition & Language

Part 2 of 4 — Problem Solving, Decision Biases & Intelligence

Two Routes to a Solution

• Algorithm: exhaustive, rule-based procedure — guarantees a solution, costs time (trying every locker combination)
• Heuristic: mental shortcut — fast, usually good enough, systematically biased in known directions
• Insight: sudden restructuring of the problem ("aha") after an impasse; incubation (stepping away) sometimes releases it

Barriers to Problem Solving

• Mental set: persisting with a previously successful strategy after it stops fitting (water-jar problems)
• Functional fixedness: seeing objects only in their usual role — Duncker's candle problem (the tack box is a shelf, once you stop seeing "container")
• Confirmation bias: searching for and weighting evidence that supports the current hypothesis; the reason good experiments try to falsify

The Judgment Heuristics (Kahneman & Tversky — know each signature error)

Related distortions: framing effects (identical options chosen differently as "90% survival" vs. "10% mortality" — people are risk-averse for gains, risk-seeking for losses); overconfidence; belief perseverance (conclusions outlive their discredited evidence); hindsight bias ("knew it all along"); gambler's fallacy (independent events "due"); sunk-cost reasoning.

Intelligence — the Theories the MCAT Contrasts

• Spearman's g: one general factor; positive correlations among all cognitive tests are its evidence
• Gardner's multiple intelligences: eight-plus independent capacities (musical, bodily-kinesthetic, interpersonal, etc.); criticized for weak psychometric separation
• Sternberg's triarchic: analytical, creative, practical
• Fluid intelligence (novel reasoning; declines with age) vs. crystallized (accumulated knowledge/vocabulary; stable or rising into later adulthood) — the standard aging dissociation
• IQ is a deviation score: mean 100, SD 15; the Flynn effect is the generational rise in raw scores (environment must matter, since gene pools cannot shift that fast)
• Heritability (proportion of variance in a population attributable to genes) is not a fixed individual quantity: it rises in enriched, uniform environments and says nothing about group differences under different conditions
• Stereotype threat: performance drops when a negative group stereotype is made salient — a situational, reversible effect that passages use to probe test-fairness interpretations

Worked Example — A Diagnostic-Reasoning Passage, Fully Read

Passage. Medical residents read a vignette of a 24-year-old with fatigue and joint pain whose presentation "closely resembles" a rare autoimmune disease (prevalence 1 in 10,000) but is also consistent with a common viral syndrome (prevalence 1 in 20). 71% of residents rank the rare disease first. In Phase 2, residents who just completed an ICU rotation featuring a fatal case of the rare disease rank it first at even higher rates. In Phase 3, giving residents explicit base-rate information cuts the error nearly in half, and requiring them to list evidence AGAINST their leading diagnosis cuts it further.

Step 1 — name the Phase 1 error. Ranking by resemblance to a memorized prototype while ignoring prevalence is the representativeness heuristic producing base-rate neglect: even a strong match to a 1-in-10,000 disease usually loses to a moderate match to a 1-in-20 one.

Step 2 — name the Phase 2 amplifier. A vivid, recent, emotionally charged case makes the rare disease easier to call to mind — availability now compounds representativeness. Note the two heuristics are dissociable: Phase 2 changed only recall ease, not the vignette's resemblance.

Step 3 — map the debiasing results. Base-rate provision attacks representativeness directly; the consider-the-opposite instruction attacks confirmation bias (residents otherwise seek features consistent with their leading hypothesis). The MCAT rewards matching each intervention to the specific bias it targets, not "education reduces bias" in general.

Step 4 — the extension item. "A resident who orders one more confirmatory test for the favored diagnosis, rather than the test that would best discriminate between the two, is displaying..." — confirmation bias, because the information sought can only support, not challenge, the current hypothesis. Discriminating tests are the clinical version of falsification.

Heuristics, Biases & Intelligence 🎯

Key Takeaways — Part 2

• Algorithms guarantee, heuristics approximate; insight restructures after impasse; mental set and functional fixedness are the two classic blocks
• Representativeness = judging by resemblance (base-rate neglect, conjunction fallacy); availability = judging by recall ease (vivid media risks); anchoring = insufficient adjustment from a starting number
• Framing: identical facts, different choices — risk-averse for gains, risk-seeking for losses; confirmation bias is countered by seeking disconfirming, discriminating evidence
• Intelligence theories: Spearman's g (positive test correlations) vs Gardner (multiple, weakly separable) vs Sternberg (analytic/creative/practical); fluid declines with age, crystallized endures
• IQ is a deviation score (mean 100, SD 15); the Flynn effect implicates environment; heritability is a population statistic, and stereotype threat is a situational depressant of scores — never an ability measure

Cognition & Language

Part 3 of 4 — Language: Structure, Acquisition & the Brain

The Structural Ladder (smallest to largest)

• English uses roughly 40 phonemes.

The Acquisition Timeline (order is what gets tested)

1. Cooing (~2 months) — vowel sounds
2. Babbling (~4-6 months) — consonant-vowel strings; initially includes ALL phonemes, universal even in deaf infants (who babble manually if exposed to sign); by ~10 months it narrows to the native language's phonemes
3. One-word (holophrastic) stage (~12 months)
4. Two-word / telegraphic speech (~18-24 months) — content words in correct ORDER, function words dropped ("want juice")
5. Overregularization (~3 years): "goed," "foots" — errors that INCREASE temporarily as rules are extracted; the child could never have imitated these forms, so they are the classic evidence against pure imitation accounts

Three Theories of Acquisition

• Behaviorist (Skinner): language learned via imitation and reinforcement — cannot explain overregularization, the poverty of the stimulus, or the speed of acquisition
• Nativist (Chomsky): an innate language acquisition device with universal grammar; supported by the critical (sensitive) period — late-rescued children (e.g., Genie) and late learners of a first sign language never reach native syntax, while vocabulary remains learnable
• Interactionist: biological readiness plus social interaction (child-directed speech, joint attention) — the modern synthesis

Language & Thought

• Whorf's linguistic relativity: language influences thought. The strong version (language determines thought) is rejected; the weak version (language nudges attention, memory, and categorization — e.g., color-term boundaries speeding discrimination) has empirical support.

Language in the Brain (lateralized, mostly left)

Global aphasia = both regions.

• The reading and writing analogs of aphasia are alexia and agraphia.

Worked Example — An Aphasia-Localization Passage, Fully Read

Passage. Three stroke patients are assessed. Patient 1 speaks in slow, strained fragments ("...son... visit... Tuesday"), follows three-step spoken commands correctly, and grows visibly frustrated. Patient 2 produces rapid, melodic sentences peppered with invented words ("the flibbers were altogether on the granting"), fails simple commands, and seems untroubled. Patient 3 converses almost normally and understands well, but cannot repeat the phrase "no ifs, ands, or buts" despite trying. A fourth finding: Patient 1's right arm is weak.

Step 1 — build the two-axis grid. Score each patient on fluency and comprehension before naming anything. P1: nonfluent + comprehension intact → Broca's (expressive) aphasia. P2: fluent + comprehension poor + neologisms + unawareness → Wernicke's (receptive) aphasia. P3: fluent + comprehends + isolated repetition failure → conduction aphasia (arcuate fasciculus).

Step 2 — use the neuro anatomy cross-checks. Broca's area sits in the left frontal lobe adjacent to the motor strip's arm/face region — P1's right-arm weakness is the confirming neighbor sign (left hemisphere controls the right body). Wernicke's area is far from motor cortex, so P2 predictably lacks weakness. The frustration/unawareness contrast follows from comprehension: P1 can hear his own errors; P2 cannot fully monitor hers.

Step 3 — pre-empt the classic distractors. "P2 has a thought disorder" — no: psychotic speech is disorganized at the IDEA level with intact grammar-word machinery, while Wernicke's speech is linguistically broken with neologisms and comprehension failure. "P3's problem is memory" — no: repetition of a just-heard phrase fails while conversation succeeds, isolating the perception-to-production relay, not storage.

Step 4 — the acquisition tie-in. If the passage adds a congenitally deaf adult who acquired sign at 20 and shows persistent nonnative sign syntax despite decades of use, that supports a critical period for GRAMMAR specifically — pair it with Chomsky, not with Skinner's reinforcement account.

Language Structure, Development & Aphasias 🎯

Key Takeaways — Part 3

• Ladder: phonemes (sound) → morphemes (meaning units) → semantics → syntax → pragmatics; grammatical nonsense proves syntax and semantics are separable
• Timeline: cooing → universal babbling that narrows by 10 months → one-word (12 mo) → telegraphic (18-24 mo) → overregularization (~3 yr, the anti-imitation evidence)
• Theories: Skinner (reinforcement — fails on overregularization and poverty of stimulus), Chomsky (LAD, universal grammar, critical period — Genie, late sign learners), interactionist synthesis; Whorf survives only in weak form
• Aphasia grid: score fluency and comprehension first — Broca's = nonfluent/comprehends/frustrated (frontal, watch for right-side weakness); Wernicke's = fluent nonsense/poor comprehension/unaware (temporal); conduction = only repetition broken (arcuate fasciculus)
• Left hemisphere dominates language in nearly all right-handers and most left-handers

• Low-yield extras: English has roughly 40 phonemes; alexia and agraphia are the reading and writing analogs of aphasia

Cognition & Language — Deep Dive

Part 4 of 4 — MCAT Integration: Consciousness, Sleep & Reading Cognition Experiments

States of Consciousness

Consciousness ranges from alert wakefulness through drowsiness, sleep, and altered states (hypnosis, meditation, drug states). Circadian rhythms (~24 h) are set by the suprachiasmatic nucleus, entrained by light, with melatonin from the pineal gland rising in darkness.

Sleep Architecture (EEG signatures are the test)

• Cycles run ~90 minutes; N3 dominates early night, REM periods lengthen toward morning
• REM rebound after deprivation implies REM serves a needed function; sleep (both SWS and REM) supports memory consolidation — the direct bridge to the learning-memory lesson
• Dream theories: activation-synthesis (cortex narrates random brainstem activity) vs. problem-solving/consolidation views; Freudian manifest/latent content is historical context

Sleep Disorders (differentiate by stage and timing)

• Insomnia (most common); narcolepsy (daytime REM intrusions, cataplexy — atonia while awake); sleep apnea (breathing stops, O2 drops, daytime fatigue; often no memory of waking)

• Narcolepsy reflects a deficiency of the wake-promoting neuropeptide orexin (hypocretin).

• Night terrors (N3, early night, no dream recall, child inconsolable) vs. nightmares (REM, later night, recalled) — a guaranteed discrimination item
• Somnambulism (sleepwalking): N3, not dream enactment

Psychoactive Drugs in One Table

Reward circuit: mesolimbic dopamine pathway (VTA → nucleus accumbens); tolerance, withdrawal, and dependence follow from adaptation to it.

Reading Cognition Experiments — the Method Layer

1. Reaction time is the workhorse dependent variable: slower RT = more processing or more conflict (Stroop); priming = faster RT
2. Within-subject designs need counterbalancing against order/practice effects; between-subject designs need random assignment
3. Check what a manipulation touched: perception, attention, memory stage, decision criterion — then match the theory whose MECHANISM lives at that stage
4. Sleep studies: self-report diaries vs. polysomnography (EEG) — objective staging beats recall; deprivation designs need a control for general fatigue

Worked Example — A Sleep-and-Memory Passage, Fully Read

Passage. Participants learn a word-pair list at 9 PM. Group A sleeps normally with polysomnography; Group B stays awake until 3 AM, then sleeps; Group C sleeps but is briefly awakened whenever slow-wave (delta) activity appears, so it gets almost no N3. At 9 AM, recall is: A 74%, B 51%, C 58%. Within Group A, recall correlates r = 0.62 with minutes of N3 sleep in the first half of the night. A second task — a mirror-tracing motor skill — shows improvement correlating instead with REM plus N2 spindle density, not N3.

Step 1 — identify the logic of three groups. B still sleeps, but later and for less time, so its deficit mixes timing and duration — it shows sleep helps but cannot say which stage matters; C sleeps a full night minus N3, isolating a stage-specific mechanism rather than sleep duration. Neither group alone would justify a stage-specific conclusion.

Step 2 — map task to stage. Declarative word pairs track N3 slow-wave sleep (early night); procedural mirror tracing tracks REM/N2 spindles. This double dissociation — two tasks, two different stage correlates — mirrors the explicit/implicit memory split: different memory systems consolidate in different sleep states.

Step 3 — flag the inference limits. The N3 correlation within Group A is correlational: deeper sleepers may differ in ways that also aid memory. The causal claim for N3 rests on Group C's selective disruption: removing N3 alone cut recall from 74% to 58%. Also note awakenings themselves stress participants — a strong answer choice will mention a yoked-control (awakened equally often at random times), and the best experimental-design option on the real exam usually is that control.

Step 4 — the integration item. "A drug that suppresses REM but preserves N3 should impair which task?" Mirror tracing (procedural), sparing word pairs — answer directly off the dissociation. If the stem instead describes morning-shifted REM loss from an early alarm, reason from architecture: REM concentrates late in the night, so early waking preferentially costs REM.

Consciousness, Sleep & Experimental Reasoning 🎯

Key Takeaways — Part 4

• EEG ladder: beta (alert wake) → alpha (relaxed wake) → N1 theta → N2 spindles/K-complexes → N3 delta (terrors, sleepwalking, early night) → REM (vivid dreams, atonia, lengthens toward morning); cycles ~90 min
• REM rebound = regulated need; declarative memory tracks N3, procedural tracks REM/N2 spindles — the sleep-consolidation double dissociation
• Disorders by stage: night terrors and somnambulism = N3 early; nightmares = REM late; narcolepsy = waking REM intrusion; apnea = breathing, not staging
• Drug classes: depressants (GABA), stimulants (dopamine/NE), opioids (endorphin receptors), hallucinogens (serotonin); reward = VTA-to-accumbens dopamine
• Experiment reading: RT indexes processing and conflict; within-subject designs demand counterbalancing; match each manipulation to the processing stage it touches before choosing a theory

• Low-yield extras: narcolepsy reflects a deficiency of orexin (hypocretin)