PSYCH240 EXAM 3

Properties of Language (CSAGD)

1. Communicative

   1. Being able to talk and converse.

2. Symbolic

   1. Language creates symbols to represent ideas.

3. (Mostly) Arbitrary

   1. What does __ mean what it does and sound the way it does? Usually not the physical sense, just the common symbol used.

      1. Exception: Kiki and Bouba experiment. “Kiki” tended to be associated with spiky shapes while “Bouba” tended to be associated with rounded shapes, indicating sound symbolism can affect how we naturally connect sounds and visual forms.

4. Generative

   1. You can produce an unlimited number of occurrences within the limits of your language structure.

5. Dynamic

   1. Constantly evolving. The english language we spoke 100 years ago is not the same english we speak now; look at current V 2014 slang, or AA V Toronto slang.

Language Levels

1. Phonemes

   1. Smallest unit of speech used to distinguish words.

   2. The atoms of sound.

   3. English has 40, other languages have more/less and different rules. (English can say “pritos“ but not “fpibs“.)

2. Morphemes

   1. Smallest unit that conveys meaning.

   2. “Psychologists” = psycho + log + ist + s

   3. Atoms of meaning.

   4. Can be entire word (eg, chair.)

   5. 50k-80k morphemes.

   6. Location matters

      1. “Steam” V “Teams”

3. Words

   1. Stored in our “mental lexicon” (brain dictionary)

   2. Combinations of 1+ morphemes

   3. English speaking adults have 100k-1mil

4. Phrases

   1. Organized groupings of words

5. Sentences

   1. Organized groupings of phrases

Syntax (+ Phrase-Structure)

Rules that determine the arrangement of words/phrases.

Phrase-structure rules: Tells you how to combine phrases to sentences. Words (usually) can’t convey meaning on their own; phrase-structure add meaning.

L/S Ambiguity

Lexical ambiguity: If a word has more than one meaning (cold, flies, etc.)

Syntactical ambiguity: Words can be grouped together in more than one phrase structure (“I saw the man with the telescope.”)

How do we process language? (bottom-up)

1. Hear sounds

2. Identify phonemes

3. Identify morphemes and words

4. Get info from words (lexical access)

5. Put words into phrases and sentences

6. Compute sentence meaning based on structure

7. See how sentence fits into the context in which it was spoken

Vision in Speech Perception (McGurk)

McGurk effect: What you see overrides what you hear. An illustration of multi-sensory integration, where conflicting visual and auditory information can lead to a different perception of a spoken phoneme. Only for speech sounds.

Language Processing is Predictive (+ cloze probability, ERPs)

1. We’re constantly predicting what is next

2. Helps us process more clearly (in the event of loud noises, etc.)

3. Some words occur more with each other than others

   1. Cloze probability: How often does the final word finish that particular sentence (high = more predictive, low = less predictive (longer reading time.))

4. Evident in brain responses measured by Event-Related Potentials (ERPs).

   1. N400: An ERP component associated with the processing of meaning and the detection of semantic anomalies.

      1. The more unlikely, the higher the N400.

   2. ERP technique: A neuroscience method used to measure electrical activity of the brain in response to stimuli, including language.

Surface V Deep Structure (in Linguistics)

S: The literal arrangement of words in a linguistic output - how the sentence is presented or spoken.

D: The underlying meaning or representation of a sentence. Theoretically, surface structures can vary widely while sharing the same deep structure.

Parsing and Garden Path Sentences

Parsing refers to the process by which sentences are analyzed into their constituent parts to understand their structure and meaning.
Garden path sentences are structured in such a way that the initial interpretation of the sentence's structure is likely to be incorrect, leading the reader to re-parse the sentence.

What Must Be Learned in Language

1. Distinguishing sounds

   1. tell regular sounds apart from speech sounds

2. Parsing language sounds

   1. parse phonemes into sounds

3. Assign meanings

   1. once you have a word, it get assigned meaning

   2. issues

      1. same sound can mean 2 different things

      2. different sounds can mean 1 thing

4. Learning rules

   1. generate grammar

   2. we don’t get grammarical feedback as a kid, just meaning feedback.

      1. correcting grammar doesn’t really have an effect on kids, they pick it up on their own eventually

Language in the Brain / Aphasia

Left side handles language.

Broca’s area (left frontal) and Wernicke’s area (superior temporal.)

Broca’s Aphasia: Slow, halting speech. Issues with production. Single grammar, no function words (the, of, be.)

Wernicke’s Aphasia: Comprehension issues. Fluent speech, but it doesn’t make any sense (made up words, word substitutes.)

Neural Pathways for Speech Comprehension and Production (how do we repeat a word)

How do we repeat a word?

1. Primary Auditory Cortex receives incoming word.

2. Wernicke’s Area processes word meaning.

3. Broca’s Area converts meaning to code for muscle movements (talking.)

4. Motor Cortex takes code and initiates muscle movements.

Development of Phonemes (think of a baby’s ability)

1st year infants can discriminate all phonemes from all languages; we gradually lose this skill when it’s not the language you’re surrounded by.

Motherese

Infant directed speech. High pitch, slow rate, exaggerated intentions, falling pitch, and pausing signals (aids parsing.) Infants prefer and respond more to motherese than regular speech.

Major Stages of Speech

1. Holophrastic Stage

   1. one word utterances

   2. no syntax, context (gestures) needed for effect

   3. under/over generalization for first ~75 words

      1. eg, dog = any small animal

   4. do understand some phrases

2. Telegraphic Stage

   1. two word utterances

   2. correct use of word order

      1. subject-action (daddy throw)

      2. action-subject (throw ball)

3. Learning Syntax Rules

   1. start learning grammar rules

   2. U-shaped curve for irregular past

      1. start using correct form (went)

      2. learn rule (add -ed) and overgeneralize it (wented)

      3. relearn correct past tense

   3. non-sense word

      1. learn general rules about applying new words

      2. 4-5 y/os know:

         1. plural of wug = wugs

         2. past tense of rick = ricked

      3. implies language learning is generative and not imitative

Children Learning Word Meaning (part/whole y shape bias)

1. When children acquire vocabulary, they do so by grasping the relationship between a part and its whole, recognizing, for instance, that a "wheel" is part of a "car".

2. Research has shown a bias toward shape, indicating that when learning nouns, children are more likely to assume that the word refers to the shape of an object rather than its color, size, or texture. This phenomenon is significant in understanding how children categorize objects and establish word meanings.

Critical Period in Language Acquisition

If you learn a language after the age of 10-12, it is highly unlikely that you will ever reach native fluency. This is seen when people learn a second language, experience social isolation, and learn sign language.

Bias toward Doer, Act, and Done-to (A>P…)

The Gleitman reading says that sentences are easier to understand when they clearly say who is doing an action and who it's done to. Active sentences, where the subject does the action, are read faster than passive ones, where the subject receives the action. But if passive sentences have clear meanings, they're understood as quickly as active ones. If a sentence can be flipped around and still make sense, its structure doesn't matter. This shows that how we organize sentences affects how we get their meaning, connecting sentence structure with understanding language.

The problem of language learning: distinguishing speech sounds, relating sounds to meaning

Language acquisition presents two main challenges.

1. People need to learn how to recognize and differentiate subtle differences in speech sounds, called phonemes.

2. They must connect these phonemes with specific meanings to communicate effectively. This is not easy, as speech sounds can vary significantly depending on the context and speaker, making the process quite complex.

Language Learning is NOT Imitation

Children actively construct language rules rather than merely imitating speech, often making grammatical and pronunciation errors. Adults typically don't directly correct these but through regular interaction and exposure, kids eventually learn correct language use.

Procedural (≠ Declarative) Information

Procedural information encapsulates the knowledge required to perform tasks and actions, which is fundamentally different from declarative knowledge that involves knowing 'that' something is the case, such as facts or events. While declarative memory allows us to recount information explicitly, procedural memory enables the execution of tasks without conscious awareness of the learned skills, such as riding a bicycle or typing on a keyboard.

Stages of Skill (CAA), Fitts & Posner

1. Cognitive

   1. declarative knowledge

   2. commit faces to memory

   3. rehearse as you try to perform

   4. requires attention

   5. may be independent of skill (describing≠doing)

2. Associative

   1. strengthen conditions which lead to a desired results

   2. feedback is vital

   3. get rid of actions that lead to errors

3. Automaticity

   1. fast

   2. executed with less attention and verbalization

   3. feedback is less important (proprioceptive>visual)

   4. consistent practice needed to maintain skill

ACT-R (Adaptive Control of Thought - Rational)

In ACT-R, people convert their knowledge (proceduralization) into production rules ('if-then' statements) for completing tasks. Initially, they see tasks as facts (declarative knowledge). Later, through knowledge compilation, they turn this knowledge into action rules (procedural knowledge). These rules become more sophisticated and effective with practice (composition), resulting in skilled performance.

Multiple Learning Systems That Operate in Parallel

Human cognition relies on multiple learning systems working simultaneously at different speeds.

1. The fast learning system quickly absorbs and uses new information, usually in a conscious and effortful way within a declarative framework. This type of learning is adaptable to new situations.

2. The slow learning system deals with implicit or procedural knowledge that is acquired gradually through repeated practice. This learning is typically unconscious and becomes automatic over time, resulting in behaviors that are harder to change once they're learned.

Fast Learning System

Characterized by the explicit or declarative acquisition of knowledge that demands conscious attention and is often tied to specific goals or tasks such as learning the capitals of countries or the rules of a game. This learning is effortful, as it requires active engagement and focus, but it also allows for the flexibility to apply the knowledge in different contexts and adapt to new information as it becomes available.

Slow Learning System

The implicit or procedural dimension of learning where the development of skills and habits occurs unconsciously through repeated practice over time. As these behaviors are rehearsed, they become automatic, demanding little to no conscious thought—evident in activities like playing a musical instrument or driving. However, this automatization also comes with a degree of inflexibility, as the learned responses are deeply ingrained and harder to alter or suppress once they have been established.

SRTT (Serial Reaction Time Test)

A cognitive experiment designed to measure implicit learning and procedural memory. Participants are asked to respond to stimuli, typically on a computer screen, by pressing buttons as quickly as possible. Over time, participants react faster, often without conscious awareness, indicating evidence of learning patterns or sequences within the task. This improvement in performance supports the existence of a slow, implicit learning system that operates beneath conscious detection.

Plus-Maze Evidence (RATz)

Hippocampus=fast system, basal ganglia=slow system. Experiment that required rats to make decisions at intersections in a plus-shaped maze. Over repeated trials, rats display improved efficiency in navigating the maze, demonstrating a form of implicit or procedural learning whereby the requisite navigational skills become automatic. Such evidence from rodents supports the theory of multiple learning systems, showing that similar to humans, animals also possess distinct mechanisms for fast, explicit learning and slower, implicit knowledge acquisition. Think HM, he couldn’t just know where something is so he would have to guess, like the rats (hippocampus). When basal ganglia is inhibited, the rats have to use place learning.

Motor Program (A&H)

Representations of our mental plan for motor movements in our brain; responsible for coordinating movements.

1. They are abstract, meaning they're not tied to specific muscles or movements. Instead, they can be adapted to control different muscles or to produce variations of a movement.

   1. Think signature demo; different motor commands for small (finger) or big (arm) signatures yet they still look the similar.

2. They are hierarchical, with the general program at the top and specific subprograms underneath. For example, when writing, your brain activates the motor program for the action, which includes subprograms for gripping the pen, forming letters, and moving your arm across the page.

   1. Think Rosembaum (typing) demo. Shows links (like TLC); the more links the longer the reaction time.

Sequential Reaction Time Evidence

In the finger-tapping experiment, participants learn a sequence of taps. As they practice, they get faster, which is evidence that they're relying on a motor program and not just reacting to each individual tap. Provides insights into how motor programs are organized. The fact that people can tap sequences faster than could be achieved by individual reactions suggests that movements are organized in a hierarchy and pre-planned.

Proprioception

The sense that lets us perceive the position and movement of our own body parts, even with our eyes closed. It's an essential aspect of motor control. Sensors in our muscles and joints send signals to our brain, letting it know where our limbs are and how they're moving. This feedback helps us coordinate our movements and maintain balance and posture. Proprioception often works without conscious thought—like when you walk without looking at your feet.

Neural Basis of Motor Control (C, BG, C)

• Cerebellum: Responsible for fine-tuning movements, balance, and coordination.

• Basal Ganglia: Involved with the initiation and regulation of motor movements, procedural learning, and habits. Parkinson’s disease.

• Cortex: The motor areas of the cerebral cortex, particularly the primary motor cortex, are responsible for the direct control of voluntary movements.

Parkinson’s Disease

A neurological disorder that affects movement, causing symptoms like tremors, stiffness, and slowness of movement. It's caused by a loss of dopamine-producing neurons in the basal ganglia, particularly in an area called the substantia nigra. This affects the basal ganglia's ability to regulate movement effectively.

Motor Homunculus (body-brain map)

A visual representation of the body parts on the primary motor cortex, showing which parts of the brain control each part of the body. The size of each body part on the homunculus is not proportional to its actual physical size, but rather to the complexity of its movements—the fingers and mouth take up disproportionately more space because of their dexterity.

Reaching Neurons in Primary Motor Cortex

These neurons in the primary motor cortex become active during arm reaching movements and are thought to help in planning and initiating the movement. When you intend to reach for something, a specific population of neurons fires in the primary motor cortex, signaling the muscles to contract in the appropriate patterns to execute the reach.

Population Coding

The way the brain represents information by using the collective activity of many neurons rather than the activity of individual neurons. Rather than relying on single "point-to-point" connections for motor control or sensory input, the brain interprets the patterns of activation across a large number of neurons. This allows for a more nuanced and robust representation of information.

Speed-Accuracy Tradeoff, Fitt’s Law

The principle that as you move faster, accuracy decreases, and vice versa. Fitts's Law mathematically models this relationship, showing that as distance increases or precision demands increase, movement time also increases.

Degrees-of-Freedom Problem

The challenge that arises because there are many possible ways to move the body's joints to achieve a specific action. We tend to choose the most efficient.

Serial Order Problem (think playing instrument)

The question of how the brain determines the sequence in which components of a movement should occur. When we produce complex actions, such as speaking or playing a musical instrument, the precise order of movements is crucial. The brain must figure out the correct sequence and timing to execute these movements correctly, which constitutes the serial order problem.

Response Chaining (and why it’s not accurate)

A theory that suggests each action in a sequence triggers the next action, like a chain reaction. However, this theory faces problems because it can't account for the speed and complexity of many sequences of actions. Response chaining would imply that each movement's completion serves as the cue for the next movement. However, due to the speed at which we can perform sequences (eg, typing) and the ability to do them in reverse or change the order, this theory doesn't sufficiently explain action sequences. It also doesn't explain how we can start in the middle of a well-learned sequence.

Hierarchal Structure of Actions (+ how it accounts for jagged reaction times in tapping task)

Actions are organized in a hierarchy, with broad goals at the top and finer, more detailed movements at the bottom. This structure allows for complex sequences to be broken down into more manageable parts. For example, the act of driving can be decomposed into sub-tasks like starting the car, operating the pedals, and steering, each of which can be further divided into more detailed actions. Hierarchical theory explains that irregular reaction times in tasks like finger-tapping are due to the brain processing and planning multiple movements at each level of the hierarchy, not just one at a time.

Motor Programs Abstract Away From Muscles

Motor programs control actions at an abstract level, meaning they define the movements in terms of the goal rather than specifying which muscles should be used. This abstraction allows for adaptability. For example, if you injure your hand, you can still perform the motor program of reaching with adjustments for the injury because the program is not tied to specific muscles.

Power Law of Practice (/ shaped graph)

States that improvements in speed or performance are rapid at first but diminish over time as you continue to practice the skill. This principle describes a logarithmic relationship between practice trials and the reaction time for task completion. Essentially, the more you practice, the less time it takes to complete a task, but the rate of improvement decreases the more you practice. This is often visualized as a negatively sloped curve on a log-log graph, where performance (like reaction time or error rate) is plotted against the number of practice trials.

What is a Problem? (+ Well-Defined V Ill-Defined)

Initial State → (non-obvious Method) → Goal State

Well-Defined: Specified initial and goal state.

Ill-Defined: Unspecified initial and goal state. “What am I gonna do (career, marriage, etc.)?”

Stages of Problem Solving

1. Form a representation of the problem

2. Construct plan

3. Execute plan

4. Check; evaluate

5. Reformulate (if goal-state is not met)

Problem Space

All the possible states and moves from the problem's starting point to its solution; a conceptual map of all the routes that can be taken, with each node representing a state or milestone, and each edge representing a possible action or decision. It's vast for complex problems, and the challenge for the problem-solver is to navigate this space efficiently.

Importance of Representation of a Problem

The way we frame or think about a problem can significantly affect our ability to solve it. An effective representation helps to clarify objectives, identify constraints, and suggest strategies for solution. For instance, mathematical problems can often be represented visually, such as using a graph, which can make a solution more apparent.

Use of Analogy (focusing on surface similarity & ignoring deep similarity = bad)

Problem-solving often involves the use of analogies, yet relying solely on obvious, surface similarities can lead to misleading solutions. This approach is akin to choosing a key based on its appearance rather than its fit for a lock. To solve problems effectively, it is crucial to recognize and understand the core principles, a process that may be less intuitive but offers profound insights. Essentially, success hinges on discerning the deeper fit between the key (solution) and the lock (problem) rather than their superficial traits.

Hinderances in Representation (TDp & FF)

1. Top-down preconceptions: Assumptions or established beliefs that can block us from seeing all possible solutions to a problem. Our previous experiences and knowledge can create a mental filter that obscures alternative perspectives or novel solutions to a problem. These top-down preconceptions can trap us in familiar patterns of thinking, making it difficult to approach a problem with an open mind. Stuck in set: the things you already know about a problem can inhibit you from finding new solutions.

2. Functional Fixedness: The tendency to see objects as functioning only in their usual or customary way; this cognitive bias limits our ability to repurpose objects or ideas to solve new problems because we're stuck in conventional uses or associations. For example, in the candle problem, individuals may fail to see a box as a potential shelf because they are fixed on its typical function of holding tacks.

Luchin’s Water Jar Task

Participants are given a problem-solving task involving three water jars with different capacities. They must figure out how to measure an exact amount of water using these jars. After solving several problems with the same solution pattern, a simpler solution emerges but many participants fail to see it because they are stuck in the previous set (mental rut) of solutions.

Algorithms V Heuristics

Algorithms are systematic, logical rules or procedures that lead to a definite conclusion.

Heuristics are mental shortcuts or educated guesses that simplify decision making. They are faster than algorithms but can be less reliable.

Think Aloud Protocol

Method of understanding cognitive processes by asking participants to speak out loud what they are thinking while performing a task. It's used to study problem-solving, learning, and cognition by providing insight into the subject's mental processing. (Eg; narrating everything you do while you cook, explaining each step so someone can understand what's happening inside your head.)

Difference Reduction Heuristic

Choosing the operator that will decrease the distance between the current situation and the desired goal state. It's often effective but can sometimes lead to a dead end. Focuses on making continuous adjustments to the current state without necessarily structuring the problem into smaller components.

Means-End Analysis Heuristic

Problem-solving strategy that involves creating subgoals (subgoaling) that will, step by step, reduce the difference between the current situation and the ultimate goal.

(A production system models human cognition as a set of condition-action rules (productions). It’s applied to means-end analysis to guide the selection and application of rules that will systematically reduce the differences between the current and desirable states until the goal is achieved.)

Hobbit and Orcs Problem

A problem-solving exercise in which you must transport groups with conflicting interests (hobbits and orcs) across a river without ever allowing a majority of orcs over hobbits on either side of the river, as this would result in the orcs eating the hobbits. (Could not be solved with difference reduction.)

Working Backwards Heuristic

A strategy involves starting from the desired end state and working in reverse order to achieve the initial conditions. It's particularly useful for problems with well-defined goals but unclear starting points.

Expertise / Chess Example

Being an expert is usually beneficial (more experience/practice, better representations), but in specific circumstances it can hinder you (functional fixedness, water jug ‘mental set’.)

In the chess example, a group of expert and novice chess players attempted to memorize two boards: one with a real game and one with a random game. Experts could remember real game better than novices, but novices remembers random boards better. (The experts knew chess patterns and where some pieces can/cannot be, so they naturally put some pieces where they weren’t when trying to remember the random board.)

10 Years to Reach World-Class Expertise

This rule originates from research that suggests it takes roughly 10 years or 10,000 hours of dedicated practice to achieve world-class expertise in any domain. This intense practice must be deliberate, targeted, and accompanied by feedback. The process follows the power law of practice, which states that as you gain more practice, the rate of improvement decreases exponentially.

Expertise: Rich Organized Schemas

Experts in a field have accumulated a wealth of knowledge, which is organized within their long-term memory in a hierarchical framework known as a schema. This rich schema allows them to quickly recognize patterns and categorize problems, based on their knowledge and past experience. For experts, these schemas are highly developed in their specific domain of expertise, allowing for swift retrieval of relevant information when facing a new problem. Can also hinder.

Expertise: Sophisticated Representations

When experts approach a problem, they create complex mental representations of it. These are "sophisticated" in the sense that they go beyond the superficial details and penetrate to the deeper structural features of the problem. Their understanding of the core principles involved allows experts to identify what is essential and what is peripheral, thus streamlining the problem-solving process.

Expertise: Spending More Time on a Problem Representation

(Longer to take in problem, quicker to solve.) Experts tend to spend a significant amount of time initially understanding and framing the problem correctly. This upfront investment in deeply representing the problem allows them to apply their extensive knowledge more effectively and discern the most appropriate approach to solving the problem.

Expertise: Less Reliance on Means-End Analysis

While means-end analysis involves comparing the current state to the goal state and then taking steps to reduce the difference, experts often rely less on this problem-solving method. Because of their rich knowledge base, they can frequently see a pathway to the solution without the need for an iterative step-by-step approach typically used by novices.

Expertise: Moving Forward with Solutions

Due to their deep understanding, experts are more likely to progress in a forward direction toward solving the problem without as much trial and error or backtracking as novices. It may take them slightly longer to process the question, but they complete a problem much faster. Their ability to select and apply the right strategy from the beginning usually leads to more efficient problem-solving.

Expertise: More Reliance on LTM

Experts have a vast reservoir of domain-specific knowledge stored in long-term memory that they can draw upon. This extensive and readily accessible knowledge base means they can quickly bring relevant information to bear on a problem without having to rely on working memory to hold and manipulate new information or develop new strategies.

Expertise: Less Reliance on WM

Because of their ability to draw on long-term memory, experts don't need to rely heavily on working memory, which is limited in capacity. They do not need to keep as many elements of the problem in conscious processing because the patterns and strategies are deeply engrained and can be recalled as a whole from long-term memory, allowing them to allocate their cognitive resources more efficiently.

Deterministic/Deductive Reasoning

(General to specific, theory → hypothesis → observation → confirmation.)

Logic where the conclusion is a necessary consequence of the premises. If the premises are true, the conclusion must be true as well. An example of deductive reasoning is a mathematical proof or using a rule of logic (e.g., syllogisms). It's analogous to a cause-and-effect mechanism; if the initial conditions are set, the outcome is predictable and inevitable.

Probablistic/Inductive Reasoning

(Specific to general, observation → pattern → hypothesis → theory.)

Reasoning where the conclusion is not guaranteed but is likely, given the premises. This is often used in everyday decision-making where certainties are rare. Inductive reasoning involves generalizing from specific instances, making predictions based on observations, and includes a degree of uncertainty. It's like forming a hypothesis based on multiple experiments or observations, which seems likely but is not absolutely certain.

Normative V Descriptive Theories

N: How one ought to be taught; ‘rules of logic’. These theories prescribe how decisions should be made to be rational or optimal. They are based on logical consistency and mathematical principles, setting standards for correct judgment and decision-making. It's equivalent to an 'ideal' model of thinking, where logic prevails, and cognitive biases are absent. In decision-making, expected utility theory is a good example of a normative model. Bayes Formula.

D: How people actually reason; biases and heuristics. These theories explain how decisions are actually made, including all the cognitive biases and errors people might exhibit. They account for the limitations of human information processing and the heuristics (mental shortcuts) people use. They are like the 'real-world' models that map unto actual human behavior, often flawed and inconsistent with the norms prescribed by logic and probability.

Normative Inductive Reasoning: Base Rate and Current Evidence (Mammogram Example)

Base Rate (p(H)): The base rate is the prior probability of an event or hypothesis. It's the general likelihood of an occurrence before any additional evidence is factored in. For example, the prevalence of breast cancer in a population would be the base rate in the context of mammogram screenings.

Current Evidence (p(E|H) and p(E|not H)): This is the evidence obtained that should be weighed against the base rate. In medical testing, this would be the reliability of the test itself – the probability that the test correctly identifies the condition (true positive rate, p(E|H)) and the probability that the test correctly identifies the absence of the condition (true negative rate, p(E|not H)).

Mammogram Example Confusion: This is a cognitive mistake where medics mistake the likelihood that a positive mammogram correctly indicates breast cancer (p(H|E)) with the likelihood that someone with breast cancer will test positive (p(E|H)). It’s a common misunderstanding rooted in difficulties with probabilistic reasoning.

Bayes’ Theorem

This is a fundamental theorem in probability theory that allows for the updating of probabilities based on new evidence. It combines the prior probability (base rate), the likelihood of observing the evidence if the hypothesis is true, and the likelihood of observing the evidence if the hypothesis is false into a posterior probability – the updated chance of the hypothesis being true having considered the new evidence.

Insufficient Weight to Current Data (50/50 base rate poker chip examples)

A phenomenon where people sometimes disregard the current data or evidence (e.g., the actual distribution of red and blue poker chips being drawn from a bowl) and rely instead on an assumed base rate (e.g., a 50/50 chance of drawing a chip of either color). Even when new data suggests the base rate is incorrect, people may not adequately adjust their beliefs.

Representativeness Heuristic (+ examples)

Tendency to judge the probability of an event by how much it resembles what we consider to be a typical example of that event. When using this heuristic, people often fail to consider the actual statistical base rate, or frequency, of the event, and instead place too much emphasis on the similarity between the event and its perceived category

• Assuming someone is more likely to be a librarian than a farmer because they are quiet and enjoy reading,

• Thinking firstborn children are more responsible because it fits the stereotype, without considering actual birth order statistics,

• Believing that after flipping several heads in a row, a tail is "due" because it fits the perception of a 50/50 event, but statistically, each flip is independent of the last.

How Representativeness Explains Doctor's Tendency to Confuse p(H|E) and p(E|H)

Doctors may confuse the probability of a hypothesis given evidence (p(H|E)) with the probability of evidence given a hypothesis (p(E|H)). This can happen because the similarity of symptoms to a disease (representativeness) may lead doctors to overestimate the likelihood of the disease given the symptoms, without considering the actual base rate of the disease.

Availability Heuristic (+ examples)

When people judge the probability of an event by the ease with which instances or associations can be brought to mind. This can lead to errors if memorable events are mistakenly believed to be more common than they actually are.

• Famous people examples: 2 groups (a: 20 famous women and 20 less famous men, b: 20 famous men and 20 less famous women.) Recall is better for famous names, groups think there are more of the famous gender.

• Subjects given category: when people are asked “how many do you think you could name in 2 minutes (with about 7 seconds to answer)?” and “name as many _ as you can in 2 minutes?”, results typically are similar.

• Death example: When people have to guess which area has more deaths (stroke V all accidents combined) they typically say the one where they have more instances of in their memory (so they may pick all accidents) even though those instances might be more memorable because it is less common.

• When partners are asked how much of the chores they do in their household, both say 80% (impossible). This is because people tend to answer based on what is in their memory; easier to recall things they know they did versus things their spouse may have done.

Base Rates

The basic, underlying rate at which an event occurs within a given population. In the context of decision-making, paying insufficient attention to base rates can lead to errors in judgment.

Illusory Correlations

The perception of a relationship between two variables when no such relationship exists, often influenced by the availability of vivid instances. (Imagine you have a friend who gets sick every time they eat oranges. You might quickly assume that oranges cause your friend to get sick. However, your friend also happens to eat oranges often during the winter, which is flu season.)

Positive Testing Strategy (Confirmation Bias)

The tendency to search for, interpret, or remember information in a way that confirms one's preconceptions, leading to statistical errors.

Gambler’s Fallacy

The incorrect belief that future probabilities are altered by past events, like expecting a coin toss to result in heads after several consecutive tails, even though the events are independent.

Hindsight Bias

The inclination, after an event has occurred, to see the event as having been predictable, even though there was no objective basis for predicting it.

Anchoring-and-Adjustment

A cognitive bias where an individual relies too heavily on an initial piece of information (the "anchor") when making decisions, and then inadequately adjusts from that starting point.

Framing Effects

People's decisions are influenced by the way a problem is presented. For example, individuals tend to be more risk-averse when a problem is framed positively (emphasizing gains) and more willing to take risks when a problem is framed negatively (emphasizing losses).

Can also influence medical decisions. For example, when a treatment is framed in terms of the number of lives saved (positive frame), people prefer it over a treatment framed in terms of the number of lives lost (negative frame), even if the statistical outcomes are the same. Disease story is an example.

Simulation Heuristic

The ease of imagining an event can influence people's judgments of the likelihood or cause of the event. This explains why scenarios that are easier to visualize are often considered more probable.

Conjunction Fallacy (feminist bank teller)

When people incorrectly believe that a specific set of conditions is more probable than a single general one. This error often arises from representativeness, where the detailed condition seems more representative of what one might expect to happen, or the availability heuristic or causal reasoning when constructing a scenario seems easier. (Linda, the feminist bank teller example.)

Why Use Non-Normative Heuristics

Non-normative heuristics may not always yield logically correct answers, but they often work well in everyday situations and require less mental resource investment. They are "fast and frugal," allowing for reasonably effective decision-making in complex, uncertain environments. These heuristics help people to make quick judgments and decisions without the need to deliberate extensively or analyze all possible outcomes.

3 Types of Deductive Reasoning (QCC)

1. Quantifier (Categorical)

   1. Involves two premises and a conclusion, each stating universal, particular, or negative relationships between categories. These statements feature major, minor, and middle terms. Validity relies on logical structure, not specific content. A frequent error is the illicit major or minor, where a term in the conclusion is distributed more broadly than in the premises.

   2. An example of categorical reasoning:

      1. Premise A: All mammals are vertebrates.

      2. Premise B: All dogs are mammals.

      3. Conclusion: Therefore, all dogs are vertebrates.

2. Comparative (Linear)

   1. Involves arranging items or individuals in a specific order based on a given attribute, such as size or age, through a series of statements. These statements may involve binary comparisons ("A is taller than B") or multi-term series ("A is taller than B; B is taller than C"). Inferences are drawn to establish a complete ordering. This form of deductive reasoning is often more intuitive, reflecting everyday experiences like arranging objects or people by size. Hierarchy.

   2. An example of linear reasoning:

      1. Premise A: Alice is taller than Bob.

      2. Premise B: Bob is taller than Charles.

      3. Conclusion: Therefore, Alice is taller than Charles.

3. Conditional (think Wason Selection Task)

   1. A form of logical reasoning that involves hypotheses and conclusions expressed through conditional statements, which assert that one condition leads to another. This form of reasoning is fundamental to deductive logic and is often represented by the "if...then..." format.

   2. An example of conditional reasoning:

      1. Premise A: If you write a good paper, you will get funded.

      2. Premise B: You wrote a good paper.

      3. Conclusion: You will get funded.

The Wason Selection Task

A logical puzzle that serves as an experimental test for conditional reasoning. In this task, participants are presented with a conditional statement and four cards. Each card has information on both sides—one that could potentially confirm or disconfirm the antecedent (P) and one that could confirm or disconfirm the consequent (Q). Participants are asked to select the card(s) that must be turned over to determine whether the conditional statement is true.

For example:

• Premise: If a card has a vowel on one side, then it has an even number on the other side.

• Given four cards showing A, K, 4, and 7, the task is to select the card(s) that must be turned over to test the truth of the statement.

Syllogisms (+VCACF)

A sequence of -usually two- statements and a conclusion, where the reader has to decide if the conclusion is true based on the statements, not on real world knowledge (eg, An apple is a fruit. All fruit is good. Therefore apples are good). Syllogisms can be inherently difficult as they require understanding and applying logical rules.

1. Validity: Some syllogisms are valid (the conclusion logically follows from the premises), others are not. People are more likely to accept valid than invalid arguments.

2. Content: Familiar content can make syllogisms easier to understand and solve.

3. Atmosphere: The use of certain quantifiers can influence how we perceive the syllogism's validity.

4. Conversion: Mistakenly inferring that the logic works both ways (e.g., "All A are B" means "All B are A").

5. Figural Effects: The position of terms can affect how easy a syllogism is to solve.

Three Theories of Deductive Reasoning: Formal, Mental, Verbal (definition, errors, and problems)

1. Formal Rule Theory

   1. Humans utilize learned or innate formal rules for solving deductive reasoning tasks. According to this perspective, individuals abstract problem content and apply these formal rules to derive conclusions.

   2. Errors occur because of misinterpreted premises, some rules are unavailable, and couldn’t find proof for conclusion.

   3. Problems

      1. If deduction relies on formal (content-free) rules, then why does content matter?

      2. Why would formal rules be built in when they only apply to infrequent tasks?

2. Mental Model Theory

   1. People construct mental models that represent premises, describe them, and then try to falsify the conclusion by constructing alternative models. (“Can I think of a world without counter examples?)

   2. Errors arise from WM limits (can’t hold on to multiple models)

   3. Problem

      1. Constructing counter examples is a very complex task that is specific to deduction. Would subjects with no training use it?

3. Verbal Reasoning Theory

   1. When people reason deductively, they work with language itself instead of abstract logical forms or mental simulations. In this view, reasoning is like processing language, where the structure and meaning of the words in the premises influence how conclusions are drawn.

   2. Errors arise because linguistic processes are adapted to communication, not deduction. In communication, we construct plausible representation of the gist (which may include inferred information that isn’t deductively valid.)

   3. Problems

      1. Doesn't clearly explain why and when language affects reasoning. It also doesn't clarify how we reason without words or understand the role of visual thinking in reasoning.

Competence V Performance Theories (The Paradox of Rationality)

• The Paradox of Rationality:

  • Competence theories focus on what cognitive processes are theoretically capable of computing, assuming optimal performance under ideal conditions.

  • Performance theories focus on how cognitive processes actually compute in practice, often under suboptimal conditions with limitations.

• The paradox arises from how people manage to navigate life successfully even though they frequently perform poorly on basic deductive reasoning tasks. The implication is that real-world rationality cannot be solely based on the ability to solve such problems.

• Performance theories suggest deductions happen in ways that include:

  1. Using content-specific rules informed by factual knowledge.

  2. Following a formal, syntactic approach that doesn't depend on content.

  3. Using semantic mental models, where a conclusion is verified across all conceivable scenarios for a situation.

Tasks and Phenomena in Deductive Reasoning (if then-MP/MT, relation reasoning, and syllogisms)

• Conditional Reasoning (If...then...):

  • Statements that depend on each other; “if it’s raining I will carry an umbrella.“

  • Valid Forms

    • Modus Ponens: "If it rains, I'll use an umbrella. It's raining, so I must be using an umbrella." It's a logical way of thinking that follows the if...then pattern correctly.

    • Modus Tollens: A bit like reverse thinking. "If it rains, I'll use an umbrella. I'm not using an umbrella, so it can't be raining." Sticks to the logical pattern, just in the opposite direction.

  • Invalid Forms

    • Affirming the Consequent: Invalid form of Modus Ponens. "If it rains, I'll use an umbrella. I'm using an umbrella, so it must be raining." But this isn't always true because you might be using an umbrella for another reason, like shading yourself from the sun.

    • Denying the Antecedent: Invalid form of Modus Tollens. "If it rains, I'll use an umbrella. It's not raining, so I'm not using an umbrella." This ignores other reasons you might have for using an umbrella.

• Reasoning about Relations

  • Understanding how different concepts are related to one another.

  • Model Theory

    • When we reason about relationships, sometimes we have to consider many different scenarios (models) to figure out what's true. If there are too many possibilities, it can be hard to decide which one is correct.

• Syllogisms and Quantifiers

  • Logical arguments that use statements to arrive at a conclusion.

  • Model Theory

    • A model-based account means you make a mental picture or story (model), decide what you think is true (generate a conclusion), and then think about other possible stories to see if your conclusion might be wrong (testing against alternative models).

Effects of Content on Deduction

We don't just use cold, hard logic. What we think is true can be swayed by what we believe or know about the world. If a logical conclusion makes sense with what we understand, we're more likely to believe it, even if another conclusion is technically possible.

Selection Task (effects of content)

This is a task where you're given a rule and have to check if certain examples follow that rule.

Content can really change how well people do on this task. If the rule is about something familiar, like a law you know about, you can use your practical knowledge (pragmatic reasoning schemas), which helps you think through the examples in a way that makes sense in the real world, not just in logic land.