Collaborative Study Guide - COGS 001.docx

Directions: With your group, utilize the slides from that topic to note key concepts and key people (might be a scholar or a person representative of a case study) Also, note in a single sentence the main contribution of each of the readings/media from that topic Computation in Cognitive Scie...

Directions: With your group, utilize the slides from that topic to note key concepts and key people (might be a scholar or a person representative of a case study) Also, note in a single sentence the main contribution of each of the readings/media from that topic Computation in Cognitive Science (Group Carey) Key concepts: Modularity Domain specificity: A module can only act on a certain type of input Mandatory processing: Module will always process its input automatically Information encapsulation: Higher-level processes can’t change how the module functions Four foundational principles Decomposition: breaking down complicated process into simple steps Materialism (mid is a product of material)/Naturalism (no spooky stuff) Reduction: going from one level to another (mental → physical) Computation: THe mind can be described in terms of information processing Rationalists VS. Empiricists Turing Machine–physical system that operates on informational input Marr’s Levels of Analysis Computational: Highest level of analysis, identifying the problem that you’re trying to solve. (ex. Vision: how to use light to determine the world’s physical structure) Algorithmic (aka Representational): What process are you using to solve the problem? (ex. To achieve vision, light gets transduced into neural firing ratings, then get interpret.ed by the visual cortex) Implementational: How the algorithm gets physically implemented, the “hardware” (ex. In Vision it’s nerves, cones/rods, etc) Operant Conditioning and reinforcement Uses stimuli and rewards to create reinforced behaviors Sounds right, but we are too complex Key people: BF Skinner (operant conditioning) - behaviorism through reinforcement learning David Marr (levels of analysis) Jerry A. Fodor (modularity of mind) Alan Turing (created a computational system) Carandini: Neural computation is an intermediate process that may be helpful in connecting the lower-level processes of neural circuits to the higher-level process of behavior. Crane: A Turing Machine: Cobb: They’re made of meat!: Everything that the mind can do can be reduced to the physical brain What happens when minds break? (Groups Gelman, Gopnik) Key concepts: Double Dissociation Visual Agnosia Prosopagnosia Can recognize objects but not faces Balint's Syndrome: Simultagnosia Can only perceive one object at a time Phantom Limb Syndrome Multi-Store memory model Sensory cortical homunculus cortex Different parts of body have sensory regions are disproportionately represented in parts of the cortex Key people: Phineas Gage construction/railway worker who was pierced in the frontal lobe by an iron rod His behavior changed after the accident: he was short tempered, inappropriate behavior (potentially compromised morals), had trouble sticking to plans, different social behaviors This showed damage to the frontal lobe may alter personality, emotions, morals, and social interactions Average Hanover Inn employee Sacks (Speed): Famous neurologist who wrote popular science books and pieces Speed talks about time perception Wernicke's (fluent) aphasia Proper syntax, devoid of meaning Broca’s (non-fluent) aphasia Devoid of syntax, proper meaning Rafal: Balint Syndrome: Simultagnosia: one object at a time Spatial Disorientation: can’t look at or reach for objects Sacks (Fresh Air): Drugs and Time Perception Clive and amnesia: Prosopagnosia ‘Test’: Evolution and nature vs. nurture (Group Hopper) Key concepts: Modularity Principles of Fodorian modules Domain specific, mandatory processing, informationally encapsulated Stroop Effect (mandatory processing) Cheater detection module Bottom up / Top down McGurk Effect bottom up = data-driven, low level top down = category-based experience, high-level Evolutionary psychology Marr’s level of analysis Computation, algorithmic, implementation Biological determinism Phrenology Reinforcement learning Operant conditioning Critical period Key people: Skinner, Chomsky, Fodor, Marr, Lamarck, Darwin Buckner: Brain is a Bag of Tricks The left side of your brain controls language, right side controls objects/motor skills. Your brain is able to multitask without your knowledge. Sugita: Face Perception of Monkeys Critical period for facial recognition Face perception is experience dependent Monkeys deprived of seeing monkey faces had more trouble differentiating between monkey faces, when deprived of human faces had trouble differentiating human faces Lewontin: The Fallacy of Biological Determinism Heritable does NOT mean unchangeable No one has ever been able to find the norm of reaction for any trait, let alone intelligence. In other words, we haven’t figured out the exact environmental conditions that change traits like intelligence exactly How you see the world (Group Kratzer) Key concepts: Perception of light, coincidence avoidance, depth perception Key people: Marr: There are three levels of visual perception: Computational, Algorithmic, Implementational. Background on color and depth perception: rods and cones- color Sacks: Music video: Really cool optical illusion etc. etc. New & Scholl: Inverse Problem: The process of inferring the cause from the effects (this is why it called "inverse": from effects to cause, rather than from cause to effects, which would be "forward"). In the case of color perception, this amounts to figuring out the "true" color of an object (i.e., the "cause") from limited information available on the retina (three types of cones reacting to different wavelengths of light). Attention and memory (Group Rosch) Key concepts: Exogenous vs Endogenous Attention: Exo = attention capture, bottom-up, externally-driven, PASSIVE (dot on screen changing to red is exogenous because we didn’t do anything to change it) Endo = volitional shifts of attention, top-down processing, ACTIVE (so you choose), and endo is bad at noticing changes… Subliminal messaging: messages passed to the mind without the mind being consciously aware of it Limited capacity theories Late selection- domain-general, Evidence against: attention affects processing of low-level features Early selection- domain-specific, modular, semantic processing only after passing through bottleneck Evidence against: semantic processing without awareness Inattentional blindness Theory of memory Can learn things without conscious memory (Clive), read words without conscious awareness Key people: Anne Treisman (late selection), William James (“My experience is what I agree to attend to”, “implies withdrawal from some things in order to effectively deal with others”) Hyman: Cell phone usage may cause inattentional blindness even during a simple activity that should require few cognitive resources, while two people may have more success due to their combined cognitive abilities. Jiang et al (gender and sexual orientation dependent spatial attentional effect of invisible images): Hypothesis: spatial distribution of attention can be changed by presence of invisible stimulus. Test: Participants are shown a blank screen, then they are subliminally shown sexual images/pixelated images. The images are taken away and the participants are shown a gabor patch. Results: if the participants quickly look at the patch, it means their attention was caught by the invisible, sexually explicit images. Sacks: Amnesia; how amnesia impacts episodic memory Intact procedural and semantic memory; able to learn new things despite having a conscious memory Simon & Levy: Concepts and conceptual development (Group Treisman) Key concepts: Quinian bootstrapping, Weber’s law, Analog magnitude representation vs. parallel individuation of small sets, Cardinal principle, MOT(Multiple Object Tracking) Key people: Piaget, Weber, Piraha people Feigenson: infants as young as 6 months old have a spontaneous ability to represent and compare quantities of objects. Results show infants have a basic ability to represent and compare small quantities, but their ability to do so declines as the numbers get larger. The experiment involved infants choosing between crackers, and the results showed they relied on object-file representations. Carey: numerical representations are a fundamental aspect of core cognition and play a crucial role in various domains of human life. There are two distinct core cognition systems with numerical content, each with its own signatures and each representing numbers in quite different ways. Understanding these systems allow us to find out how children come to quantify over sets using the conceptual resources that underlie natural language quantifiers and how they learn the meanings of explicit numerals. Learning a language (Group Churchland) Key concepts: Nativism: nature Empiricism: nurture Semantics: the meaning of langues vs. syntax: the structure of language Linguistics: Phonology: Study of combination of sounds , morphology: smallest unit that has meaning Computer cognition: how do we train models Key people: Skinner: Nurture (Behavioralist Approach) Pigeon Experiment: Teach a pigeon to respond to specific simuli using a red and green light that indicates food availability Pavlov’s Dog Experiment Noam Chomsky: Poverty of Stimuli: There is too much stimuli in human language and living experience in general to map any one stimuli to a specific behavior or verbal response Disputes Skinner's Behavioralist theory specifically Disconnect between syntax and semanitcs Readings Friedmann: Humans have a critical period for developing language which is during their first year of life. McCulloch: We won’t speak emoji because of lack of discrete combinatorial systems (syntax). However, emojis do have lexicon (arbitrary signs). Kuhl: Infants map critical aspects of language in the first year of their life. Statistical properties are picked up through exposure to the language. Understanding and language (Group Kanwisher) Key concepts: Operant Condition Arbitary Symbols (lexicon) Discrete Combinatorial Systems (Syntax) Text Prediction Word2Vec Chat GPT4 Semantics Pragmatics Key people: Adger: Hofstadter: Translation as a subtle art that Google Translate is bad at Humans use their experience and imagination Google Translate doesn’t try to understand, just tries to decode Google Translates processes pieces or text, not the whole thing About translating ideas, not words Chomsky: False Promise of ChatGPT “The Crux of machine learning is description and prediction,” an explanation is descriptions + predictions + the ability to think in alternative possibilities for past & future events + additional clauses (cause & effect, etc) ChatGPT is incapable of distinguishing between impossible and possible, they only memorize things (they can “learn” that the Earth is flat), humans learn through trial and error + we have universal grammer To be intelligent, one needs to be able to make moral decisions, which ChatGPT cannot do, “moral indifference bornn of unintelligence” The wise words of Nim Chimpsky according to Wikipedia: “Three-sign quotations Apple me eat Banana Nim eat Banana me eat Drink me Nim Eat Nim eat Eat Nim me Eat me Nim Eat me eat Finish hug Nim Give me eat Grape eat Nim Hug me Nim Me Nim eat Me more eat More eat Nim Nut Nim nut Play me Nim Tickle me Nim Tickle me eat Yogurt Nim eat Four-sign quotations Banana Nim banana Nim Banana eat me Nim Banana me Nim me Banana me eat banana Drink Nim drink Nim Drink eat drink eat Drink eat me Nim Eat Nim eat Nim Eat drink eat drink Eat grape eat Nim Eat me Nim drink Grape eat Nim eat Grape eat me Nim Me Nim eat me Me eat drink more Me eat me eat Me gum me gum Nim eat Nim eat Play me Nim play Tickle me Nim play Longest recorded quotation Nim's longest "sentence" was the 16-word-long "Give orange me give eat orange me eat orange give me eat orange give me you." “