Quiz 4 v2

Semantic networks represent knowledge as nodes (concepts) connected by labeled relationships (e.g., 'is-a', 'has-a'). They enable context-sensitive retrieval through **spreading activation**: activating one node spreads activation to related nodes, making associated concepts more accessible — explaining priming effects and context-dependent perception.

Mental models are internal simulations of specific situations used to reason and problem-solve, rather than applying abstract logical rules. They explain reasoning errors (e.g., failures in syllogistic logic) because people reason from constructed scenarios — making them more psychologically realistic but also error-prone and cognitively demanding.

Mental models are dynamic simulations built for specific, novel situations — flexible but cognitively costly. Schemas/scripts/frames are static stored structures for familiar categories or recurring events — faster to apply but less adaptable. Mental models are more useful when the situation doesn't match a stored structure.

WordNet organizes words into synsets (synonym sets) linked by semantic relations (hypernymy = 'is-a', meronymy = 'part-of', etc.), mirroring how humans store and relate words in the mental lexicon. It bridges cognitive linguistics and AI by making semantic relationships computable.

The mental lexicon is the cognitive system storing a person's vocabulary — word meanings, pronunciations, grammatical roles, and semantic relationships. It is organized by both semantic connections (related meanings) and phonological connections (similar sounds), functioning as an internal dictionary rather than an alphabetical list.

Spreading activation is the process by which activating one concept node automatically spreads activation to related nodes. It explains **semantic priming**: e.g., seeing 'doctor' speeds up recognition of 'nurse' because activation spreads along the stored association before the second word appears.

In the **preattentive stage**, simple features (color, orientation, motion) are detected in parallel across the visual field — search time is fast and does not increase with display size. Combining features (conjunctive search) requires focused serial attention, so search time increases linearly with the number of distractors. This explains pop-out vs. conjunction search differences.

Working Memory (short-term) has limited capacity (~7±2 items) and brief duration (~20–30 sec without rehearsal); it acts as a gateway to Long-Term Memory (unlimited capacity, relatively permanent). Rehearsal transfers information into LTM — without it, information decays and is lost, directly explaining everyday forgetting.

Miller's 'Magic Number 7±2': STM holds ~7 chunks simultaneously. Capacity is increased through **chunking** — grouping individual items into meaningful units (e.g., remembering 'FBI' as one chunk instead of three letters), effectively expanding the information content per slot.

• **Atkinson-Shiffrin (Modal):** Sensory → STM → LTM; rehearsal transfers info; linear flow • **Baddeley's Working Memory:** Replaces STM with Central Executive + Phonological Loop + Visuospatial Sketchpad + Episodic Buffer; emphasizes active manipulation • **Anderson's ACT-R:** Declarative (chunks) + Procedural (if-then rules); memory as activation levels that decay

• **Iconic** (sensory visual): ~250–500 ms • **Echoic** (sensory auditory): ~3–4 seconds • **Short-Term Memory:** ~15–30 seconds without rehearsal • **Long-Term Memory:** potentially permanent (minutes to lifetime)

Both are **explicit/declarative** memories. • **Episodic:** personally experienced events tied to time and place (e.g., your first day of school) • **Semantic:** general world knowledge not tied to personal experience (e.g., knowing Paris is the capital of France)