Cognition Study Guide

📖 Core Concepts Cognition – mental processes that acquire, transform, store, retrieve, and use information; includes perception, attention, memory, language, decision‑making. Controlled vs. Automatic – controlled processes are intentional, flexible, resource‑heavy; automatic processes are effortless, unconscious, and become automatic with practice. Memory Systems – working memory (short‑term manipulation), long‑term memory (episodic, semantic, procedural). Reasoning Types – Deduction (certainty if premises true), Induction (probable generalizations), Abduction (best explanation). Dual‑Process Theory – System 1 = fast, intuitive, automatic; System 2 = slow, deliberative, controlled. Cognitive Load – intrinsic (task complexity), extraneous (poor design), germane (learning‑promoting). Computationalism vs. Connectionism – symbol manipulation vs. parallel distributed networks of weighted nodes. 4E Cognition – embodied, embedded, extended, enactive – cognition is not just internal representation. --- 📌 Must Remember Controlled → Automatic: Repeated practice can shift a process from controlled to automatic. Expected Utility: \(\text{EU} = \sum{i} pi \times vi\) (maximize EU for rational decisions). Memory Encoding Boosters: Attention, emotion, mood, and contextual cues enhance long‑term storage. Heuristics & Biases – Availability, Representativeness, Anchoring → systematic errors in judgment. Stages of Development (Piaget): Sensorimotor → Pre‑operational → Concrete‑operational → Formal‑operational. Metacognition – monitoring (“Do I know this?”) + regulation (“Should I reread?”) improves performance. Cognitive Load Theory – keep extraneous load low; align intrinsic load with learner expertise; promote germane load. --- 🔄 Key Processes Perceptual Encoding → feature detection → object recognition → integration with memory/expectations. Memory Formation: Acquisition → Retention (consolidation, often during sleep) → Retrieval (cue‑dependent). Problem Solving: Identify initial state → define goal state → list constraints → decompose into sub‑problems → apply heuristics (analogy, decomposition) → converge on solution. Decision Making (Expected Utility): List alternatives → assign probabilities \(pi\) → assign values \(vi\) → compute EU → choose max EU. Metacognitive Regulation: Planning → monitoring → evaluating → adjusting strategies. --- 🔍 Key Comparisons Deductive vs. Inductive Reasoning Deductive: guarantees truth if premises true; moves from general → specific. Inductive: infers general rules from specific instances; conclusions are probable. System 1 vs. System 2 System 1: fast, parallel, heuristic‑driven, low effort. System 2: slow, serial, rule‑based, high effort. Symbolic (Computationalism) vs. Connectionist Models Symbolic: discrete symbols, explicit rules, “mentalese”. Connectionist: distributed representations, learning via weight adjustment. Controlled vs. Automatic Processes Controlled: conscious, flexible, limited capacity. Automatic: unconscious, rigid, high capacity. --- ⚠️ Common Misunderstandings “All cognition is conscious.” – Many processes (e.g., face recognition) operate automatically without awareness. “Memory works like a perfect video recorder.” – Retrieval is cue‑dependent; memories are reconstructive and prone to bias. “Heuristics are always wrong.” – Heuristics are efficient shortcuts; they become biases only when misapplied. “Dual‑process means two separate brains.” – Both systems are instantiated in overlapping neural circuits; they compete for output. “Procedural memory is “remembering how” and thus conscious.” – Procedural memory is largely implicit and non‑declarative. --- 🧠 Mental Models / Intuition “Cognitive Architecture as a Factory” – Input (sensory) → Assembly line (attention, working memory) → Storage (long‑term) → Shipping (retrieval). “Biases as Filters” – Think of availability bias as a “most‑visible‑items” filter that hides less‑visible data. “Dual‑process as Two Competing Drivers” – System 1 pushes fast, System 2 applies brakes; strong brakes (high cognitive load) let System 1 dominate. --- 🚩 Exceptions & Edge Cases Automaticity Limits – Not all controlled skills become automatic (e.g., complex ethical reasoning). Memory Enhancers – Emotional arousal boosts memory only if it aligns with the encoded material; unrelated stress can impair encoding. Bias Override – High motivation and accountability can reduce anchoring but may increase overconfidence. Developmental Exceptions – Some children display early formal‑operational reasoning in specific domains (e.g., math) before age 12. --- 📍 When to Use Which Choose Symbolic Models when the task requires clear rule manipulation (e.g., logical puzzles, formal reasoning). Choose Connectionist Models for pattern recognition, language learning, or tasks with noisy data. Apply System 1 Heuristics for quick everyday judgments under time pressure. Engage System 2 for novel, high‑stakes decisions, especially when biases are identified. Use Cognitive Load Strategies (segmenting, pre‑training) when teaching complex material to novices. --- 👀 Patterns to Recognize “Cocktail‑party effect” → selective attention cue; indicates that attention can filter by relevance despite background noise. “Ill‑structured problem” → vague goal, multiple constraints → need for divergent thinking first. “Anchoring + Adjustment” → initial value strongly influences subsequent estimates; look for a presented number before a judgment. “Chunking in Working Memory” – groups of 3‑4 items are typical; larger sets indicate chunking strategies. --- 🗂️ Exam Traps Distractor: “All cognition is conscious.” – Wrong; automatic processes are unconscious. Distractor: “Procedural memory is declarative.” – Incorrect; procedural memory is implicit. Distractor: “Heuristics always lead to errors.” – Overgeneralization; many heuristics are adaptive. Distractor: “Connectionist models cannot represent rules.” – Misleading; connectionist networks can learn rule‑like patterns through training. Distractor: “Intrinsic load is the same as difficulty of content.” – Not exactly; intrinsic load also depends on element interactivity, not just difficulty. ---