Course outline
Module 1
Module 1: The Birth of Cognitive Science
The Birth of Cognitive Science
Knowledge check
Q1. Which 1956 event is most often cited as the symbolic starting point of the cognitive revolution?
- A. The publication of B.F. Skinner's 'Verbal Behavior'
- B. The Dartmouth Summer Research Project on Artificial Intelligence and the MIT Symposium on Information Theory ✓
- C. The founding of the American Psychological Association
- D. The release of the first electronic digital computer
Why: The 1956 MIT Symposium on Information Theory, alongside the Dartmouth AI workshop, brought together researchers in psychology, linguistics, computer science, and information theory, and is widely treated as the moment cognitive science began to coalesce as an interdisciplinary field.
Q2. Which school of thought dominated American psychology before the cognitive revolution and rejected the study of mental processes as unobservable?
- A. Gestalt psychology
- B. Psychoanalysis
- C. Behaviorism ✓
- D. Humanistic psychology
Why: Behaviorism, associated with John B. Watson and later B.F. Skinner, argued that psychology should study observable stimuli and responses rather than inner mental states, which it considered unscientific.
Q3. Noam Chomsky's 1959 review of B.F. Skinner's 'Verbal Behavior' was important because it:
- A. Defended behaviorism against its critics
- B. Argued that language acquisition cannot be explained by stimulus-response learning alone ✓
- C. Proposed that language is entirely learned through imitation
- D. Introduced the concept of the computer metaphor for the mind
Why: Chomsky argued that children learn languages too quickly and creatively, producing novel sentences, for reinforcement and imitation to account for it. His critique helped undermine behaviorist explanations and pushed attention toward internal mental structures.
Q4. George Miller's 1956 paper 'The Magical Number Seven, Plus or Minus Two' contributed to cognitive science by:
- A. Showing that human memory capacity for chunks of information is limited ✓
- B. Proving that rats can solve mazes using logic
- C. Demonstrating that language is innate
- D. Establishing the speed of neural transmission
Why: Miller showed that people can hold roughly seven (plus or minus two) items in short-term memory, and that chunking can expand effective capacity. It was an early example of quantitatively studying internal information-processing limits.
Q5. The 'information processing' approach that shaped early cognitive science primarily viewed the mind as analogous to:
- A. A biological organism evolving over time
- B. A digital computer manipulating symbols ✓
- C. A mirror reflecting the environment
- D. A hydraulic pressure system
Why: Early cognitive scientists such as Allen Newell and Herbert Simon treated the mind as a symbol-manipulating system, comparable to a computer running programs. This computational metaphor became central to the field's framework.
Module 2
Module 2: Information Processing Models of Mind
Information Processing Models of Mind
Knowledge check
Q1. Which metaphor best captures the core assumption of information processing models of mind?
- A. The mind as a blank slate shaped by experience
- B. The mind as a computer that receives, transforms, stores, and retrieves symbolic information
- C. The mind as a set of unconscious drives in conflict
- D. The mind as a reflex system responding to environmental stimuli
Q2. In the Atkinson-Shiffrin multi-store model, what is the role of the sensory register?
- A. To permanently store all incoming information
- B. To hold a brief, large-capacity copy of raw sensory input before it is attended to
- C. To manipulate information through active reasoning
- D. To transfer memories into long-term storage through rehearsal
Q3. What did George Miller's 1956 paper 'The Magical Number Seven, Plus or Minus Two' propose about short-term memory?
- A. Short-term memory holds roughly seven items, or chunks, at a time
- B. Short-term memory has unlimited capacity but decays within seconds
- C. Short-term memory can only store visual information
- D. Short-term memory holds exactly seven digits regardless of their meaning
Q4. Which stage of information processing is concerned with selecting which stimuli receive further processing?
- A. Encoding
- B. Storage
- C. Attention
- D. Retrieval
Q5. Why did information processing approaches favor computer analogies in studying cognition?
- A. Computers were proven to be biologically identical to brains
- B. Computers provided a framework for describing mental processes as sequences of symbol manipulation that could be modeled and tested
- C. Computers showed that mental processes are purely emotional
- D. Computers demonstrated that cognition does not require any internal representations
Module 3
Module 3: Language and the Mind
Language and the Mind
Knowledge check
Q1. Which scholar's 1959 review of B.F. Skinner's book 'Verbal Behavior' is often credited with challenging behaviorist accounts of language and helping launch the cognitive revolution?
- A. Jean Piaget
- B. Noam Chomsky ✓
- C. George Miller
- D. Ulric Neisser
Why: Noam Chomsky's review argued that behaviorist principles of stimulus, response, and reinforcement could not explain how children acquire the creative, rule-governed capacity to produce and understand novel sentences. This critique is widely cited as a turning point in the shift toward cognitive explanations of language.
Q2. What term did Chomsky use to describe the system of unconscious rules that allows speakers to generate an infinite number of grammatical sentences?
- A. Semantics
- B. Pragmatics
- C. Universal grammar ✓
- D. Phonology
Why: Chomsky proposed that humans possess an innate, species-wide language faculty, often called Universal Grammar, which constrains the possible structures of human languages. Semantics and pragmatics concern meaning and language use, while phonology concerns sound systems, so none of those describes the rule system itself.
Q3. What is the 'poverty of the stimulus' argument in language acquisition?
- A. Children learn language slowly because they receive too little adult speech to practice with
- B. The input children hear is too limited and noisy to explain how they acquire complex grammatical knowledge without innate constraints ✓
- C. Children from low-income households develop smaller vocabularies than their peers
- D. Language learning declines in adulthood because neural plasticity decreases
Why: The poverty of the stimulus argument holds that the linguistic evidence children receive is fragmentary, error-filled, and underdetermines the grammar they end up knowing. Since children reliably acquire complex rules, such as structure-dependent question formation, some built-in knowledge must guide learning. It concerns the logical adequacy of input, not the quantity of speech or socioeconomic status.
Q4. Which phenomenon shows that children actively construct grammatical rules rather than simply imitating adult speech?
- A. Babbling in the first year of life
- B. Overregularization, such as saying 'goed' instead of 'went' ✓
- C. Using gestures before speaking
- D. Recognizing their own name in speech
Why: Children who say 'goed' or 'breaked' have almost certainly never heard those forms from adults. Their errors reveal that they have extracted a general past-tense rule (add -ed) and applied it to irregular verbs. This is evidence of rule-based learning rather than mere imitation.
Q5. In the cognitive view of language, what does 'mental representation' refer to?
- A. The physical sound waves produced when a person speaks
- B. Internal structures or states in the mind that stand for linguistic information such as words, meanings, or sentence structures ✓
- C. The written transcription of spoken language
- D. The social agreements that determine word meanings within a community
Why: Cognitive science treats language processing as computation over internal representations, such as stored word forms, syntactic trees, and conceptual meanings. Sound waves and written transcripts are external signals, and social conventions are public norms, so none of those is the internal mental structure the concept refers to.
Module 4
Module 4: Memory Systems and Representation
Memory Systems and Representation
Knowledge check
Q1. Which memory system, proposed by Atkinson and Shiffrin, is characterized by a very brief retention of sensory information before it fades or is passed on for further processing?
- A. Working memory
- B. Sensory memory ✓
- C. Long-term memory
- D. Procedural memory
Why: Sensory memory holds raw sensory input for a very short time (less than a second for visual information, a few seconds for auditory). Information that receives attention moves on to short-term or working memory, while the rest decays.
Q2. What is the approximate capacity of short-term memory proposed by George Miller in his 1956 paper?
- A. 1 to 2 items
- B. 7 plus or minus 2 items ✓
- C. 15 to 20 items
- D. Unlimited items
Why: Miller's influential 'magical number seven, plus or minus two' suggested that short-term memory can hold about 7 chunks of information. Later research suggested the capacity may be closer to 4 chunks, but Miller's estimate remains a classic reference point.
Q3. Which term describes grouping individual pieces of information into larger, meaningful units to make them easier to remember?
- A. Rehearsal
- B. Encoding
- C. Chunking ✓
- D. Decay
Why: Chunking combines separate items into meaningful clusters, effectively increasing the amount of information held within limited short-term memory capacity. For example, remembering the letters F-B-I-C-I-A-I-R as FBI, CIA, IRS is easier than remembering eight separate letters.
Q4. Which type of long-term memory involves the recall of facts and events that can be consciously brought to mind?
- A. Procedural memory
- B. Implicit memory
- C. Explicit (declarative) memory ✓
- D. Sensory memory
Why: Explicit or declarative memory covers conscious recollection of facts (semantic memory) and personal experiences (episodic memory). Procedural memory, by contrast, is implicit, supporting skills like riding a bike that are performed without conscious recall of the steps.
Q5. In the cognitive revolution, what was the primary metaphor used to describe the mind's information processing?
- A. A mechanical clock
- B. A computer ✓
- C. A steam engine
- D. A mirror reflecting the world
Why: Cognitive psychologists modeled the mind as an information-processing system analogous to a computer, with inputs, storage, retrieval, and outputs. This computational metaphor drove research on how information is encoded, stored, and retrieved.
Module 5
Module 5: Attention and Perception
Attention and Perception
Knowledge check
Q1. In the dichotic listening experiments of the 1950s, what did Cherry (1953) find participants could report about the unattended message?
- A. Its full semantic content
- B. Its physical features, such as the speaker's gender and whether the sound was speech or tone
- C. Every word that was spoken
- D. Nothing at all, not even that a sound had occurred
Q2. Which model of attention proposes that selection occurs early in processing, before semantic analysis?
- A. Late selection theory
- B. Attenuation theory
- C. Early filter theory (Broadbent)
- D. Load theory
Q3. What does the term 'cocktail party effect' describe?
- A. The tendency to forget conversations after a party
- B. The ability to focus on one conversation while filtering out others in a noisy environment
- C. The inability to hear anything in crowded rooms
- D. The increase in loudness people use when talking at parties
Q4. In Treisman's attenuation theory, what happens to unattended information?
- A. It is completely blocked and never processed
- B. It is weakened but can still be processed, with important or familiar items passing a threshold
- C. It is stored permanently in long-term memory without analysis
- D. It is fully processed and then discarded from working memory
Q5. What is the main finding of the Stroop effect that supports the idea of automatic processing?
- A. People name word meanings faster than ink colors
- B. Reading words interferes with naming ink colors, even when reading is irrelevant to the task
- C. Color blindness impairs reading speed
- D. Participants cannot read words printed in color
Module 6
Module 6: Problem Solving and Reasoning
Problem Solving and Reasoning
Knowledge check
Q1. In cognitive psychology, what is the term for a mental shortcut that uses readily available examples to judge the likelihood of an event?
- A. Anchoring
- B. Availability heuristic ✓
- C. Confirmation bias
- D. Framing effect
Why: The availability heuristic leads people to estimate how likely something is based on how easily examples come to mind, such as overestimating plane crash risk after seeing news coverage of one.
Q2. Which term describes a mental representation of the steps needed to reach a goal, often described as a path through a set of possible states?
- A. Problem space ✓
- B. Schema
- C. Algorithm bias
- D. Working memory buffer
Why: Newell and Simon described problem solving as searching a problem space, which consists of the initial state, goal state, and all intermediate states reachable through legal operations.
Q3. What is the name for the tendency to fixate on the usual function of an object and fail to see a novel use for it?
- A. Functional fixedness ✓
- B. Set shifting
- C. Belief perseverance
- D. Hindsight bias
Why: Functional fixedness, studied by Duncker, is the difficulty in seeing an object as useful for a purpose other than its conventional one, which can block solutions to problems like the candle problem.
Q4. Which type of reasoning moves from general premises to a specific conclusion that must be true if the premises are true?
- A. Inductive reasoning
- B. Abductive reasoning
- C. Deductive reasoning ✓
- D. Analogical reasoning
Why: Deductive reasoning guarantees its conclusion when premises are true, as in 'All mammals breathe air; whales are mammals; therefore whales breathe air.' Inductive reasoning, by contrast, only makes conclusions probable.
Q5. What is a means-ends analysis strategy in problem solving?
- A. Randomly trying every possible solution until one works
- B. Reducing the difference between the current state and the goal state by selecting actions that move closer to the goal ✓
- C. Relying on past experience to recall an identical solved problem
- D. Dividing a problem into unrelated subproblems solved independently
Why: Means-ends analysis, developed by Newell and Simon in the General Problem Solver, identifies the difference between the current and goal states and chooses operators that reduce that difference.
Module 7
Module 7: Connectionism and Neural Networks
Connectionism and Neural Networks
Knowledge check
Q1. What is the basic processing unit in a connectionist network, loosely modeled on biological neurons?
- A. A symbol
- B. A node (unit) ✓
- C. A production rule
- D. A semantic network link label
Why: Connectionist models consist of interconnected simple processing units, or nodes, that receive, combine, and pass on activation, loosely inspired by biological neurons.
Q2. In a connectionist network, what primarily determines how strongly one unit influences another?
- A. The length of the connection in physical space
- B. The weight on the connection between them ✓
- C. The number of symbols stored in each unit
- D. The order in which units were created
Why: Connection weights encode the strength and sign (excitatory or inhibitory) of influence between units, and knowledge is largely stored in these weights.
Q3. What is the role of the learning rule in a neural network such as one trained with backpropagation?
- A. To manually program each rule into the network
- B. To adjust connection weights based on errors so performance improves ✓
- C. To delete units that are rarely activated
- D. To convert distributed patterns into explicit symbols
Why: Learning rules like backpropagation compute how much each weight contributed to output error and adjust weights to reduce that error over training.
Q4. Which property of connectionist representations means that information is spread across many units rather than stored in a single location?
- A. Localist representation
- B. Serial processing
- C. Distributed representation ✓
- D. Symbolic compositionality
Why: In distributed representations, each concept is encoded by a pattern of activity across many units, and each unit participates in representing many concepts.
Q5. Why were early single-layer perceptrons unable to learn the XOR function, and what was the historical impact of this limitation?
- A. XOR requires more memory than perceptrons had, which led to hardware advances
- B. XOR is not linearly separable, and this limitation contributed to skepticism about neural networks until multilayer networks with hidden layers became trainable ✓
- C. XOR requires symbolic logic that neural networks cannot represent at all
- D. XOR has too many inputs for perceptrons to process in real time
Why: Minsky and Papert showed single-layer perceptrons cannot solve non-linearly separable problems like XOR; this fueled reduced interest until hidden layers and backpropagation made such problems solvable.