Perception 2

consciousness (last 3), attention, memory, problem solving (1st 3)

🔍 What Is Consciousness?

• The subjective experience of being aware, sentient, and able to reflect on thoughts.

• John Searle: “Consciousness is that experience of subjective being that begins when you wake up and ends when you fall asleep.”

🧩 Features of Consciousness

1. Sentience: Sensory awareness (vision, sound, smell, etc.)

2. Wakefulness: Not in deep sleep or coma.

3. Self-awareness: Awareness of being aware.

4. Meta-cognition: Ability to think about one's own thoughts.

🧪 Is Consciousness a Scientific Problem? -points against and for

❌ Arguments Against:

• Consciousness is subjective, private, and first-person.

• It cannot be directly observed or measured like external behaviour.

• Each person’s experience is unique and inaccessible to others.

✅ Arguments For:

• Psychology is a scientific discipline aiming to explain all mental phenomena.

• Consciousness arises from physical brain states, which can be measured.

• Tools like fMRI, MEG, and multi-unit recordings allow researchers to correlate brain activity with reported experiences.

• The challenge is technical, not conceptual.

Example: fMRI scans show which brain areas activate during specific tasks (e.g., V4 for colour, V5 for motion), allowing researchers to study the neural correlates of consciousness.

2 Types of Consciousness

1. Phenomenal Consciousness

• Subjective experience — what it “feels like” to be you.

• Includes qualia (the inner phenomena eg.when u c strawberry u c red, experience redness -u have a qualia). -Cannot be directly measured or shared.

2. Reflexive Consciousness (Meta-cognition)

• Thinking about your own thoughts.

• Enables reflection, imagination, and learning.

Historical Approaches to Consciousness; Philosophy of Mind

• Descartes: “I think, therefore I am” → Dualism (mind and body are separate).

• Materialism: Mind arises from physical brain processes.

• Epiphenomenalism: Mental states are by-products of brain activity but don’t influence it.

Historical Approaches to Consciousness; Introspectionism (Late 19th Century)

• Early psychologists tried to study consciousness by self-observation.

• Focused on qualia.

• Failed due to subjectivity and lack of reproducibility.

eg.Observers described their experience of “redness,” but results varied widely.

Historical Approaches to Consciousness; 🐀 Behaviourism (Early 20th Century)

• Rejected the mind; focused only on observable behaviour.

• Psychology became stimulus-response science.

Historical Approaches to Consciousness; 💻 Cognitive Psychology (1960s)

• Brain viewed as an information processor.

• Metaphor: brain = hardware, thoughts = software.

• Still didn’t address consciousness directly.

Historical Approaches to Consciousness; 🧠 Neuroimaging (1990s Onward)

• Tools like fMRI, MEG, and multi-unit recordings allowed scientists to study the brain while people are conscious.—> Enabled objective measurement of brain activity alongside subjective reports.

The Easy vs. Hard Problem of Consciousness

Easy Problem

• Identify neural correlates of consciousness (NCC).

• Correlate brain activity with specific conscious experiences.

eg.V4 activates during colour perception; V5 during motion perception.

Hard Problem

• Explain why brain activity produces subjective experience.

• Why does neural firing result in qualia? when u look inside brain of wet neural mess somehow u get consciousness, how??

eg.Knowing which part of the brain lights up during math doesn’t explain why it feels like doing math.

What Is an Emergent Property?

An emergent property is a characteristic of a complex system that does not exist in its individual parts, but arises when those parts interact.

eg.-Water molecules aren’t wet, but water is. -Heart cells don’t pump blood, but the heart does.

• The brain’s extreme complexity (100 billion neurons, 500 trillion synapses) allows consciousness to emerge.

• It’s not found in individual neurons but in the networked system.

Embodied Consciousness: Why a Brain in a Jar Isn’t Conscious

• Consciousness requires sensory input and bodily interaction.

• A brain without a body has nothing to perceive and no way to act.

• Sensory systems are essential for being aware of the world.

eg.A brain in a jar lacks eyes, ears, and a body — so it has no input and no output. It cannot be conscious.

Why Do Humans Need Consciousness?

🔹 To Interact with the World

• Gather information.

• Make decisions.

• Plan actions.

• Learn from experience.

🔹 To Function in Social Groups

• Understand others’ thoughts and feelings (Theory of Mind).

• Share ideas and collaborate.

• Build culture and knowledge.

eg.Empathy, cooperation, and communication all require consciousness.

Binocular Rivalry

• A visual phenomenon where incompatible images are presented to each eye. The brain cannot fuse the images, so perception alternates between them. Even though both images enter the visual system, only one is consciously perceived at a time.

eg.One eye sees Trump, the other sees Xi — perception flip-flops between the two.

• It dissociates visual input from visual awareness.

• Both stimuli are present, but only one is consciously experienced.

• This allows researchers to identify Neural Correlates of Consciousness (NCC) — the minimal neural activity required for a conscious percept.

eg.Like a toggle switch in the brain — when one image is perceived, the corresponding neurons are active, and the others are suppressed.

🧠 Neural Correlates of Consciousness (NCC)

• The minimal set of neuronal events sufficient for a specific conscious experience.

• Binocular Rivalry and NCC:

  • Neurons that modulate with perception (i.e., turn on/off depending on what is consciously seen) are candidates for NCC.

  • In rivalry, if a neuron fires only when Trump is perceived and goes silent when Xi is perceived, it is likely part of the NCC for Trump.

Microstimulation

• Involves direct electrical stimulation of specific brain areas using depth electrodes.

• Allows precise mapping of perceptual experiences.

eg.Stimulating the FFA (Fusiform Face Area) causes people to perceive faces in objects — showing FFA is necessary and sufficient for face perception.

TMS (Transcranial Magnetic Stimulation):

• A non-invasive technique that temporarily disrupts or activates brain regions.

• Used to create transient lesions or evoke percepts.

eg.TMS over MT (Middle Temporal area) evokes motion phosphenes (illusory motion), showing MT is linked to motion perception.

Key Findings:

• MT and FFA are both necessary and sufficient for motion and face perception.

• V1 (Primary Visual Cortex) is also necessary — feedback to V1 is required for conscious perception. -Blindsight Evidence: Patients with V1 lesions do not perceive motion phosphenes even when MT is intact, suggesting feedback to V1 is essential.

Access Consciousness vs Phenomenal Consciousness

Access Consciousness: info that is available for use by cognitive systems (e.g., decision-making, motor control).

• Explained by GNW.

Phenomenal Consciousness: The subjective experience of being aware.

• Explained by RPT.

Global Neuronal Workspace (GNW)

• Consciousness arises when information enters a global workspace in the brain. This workspace is accessible to multiple systems: memory, attention, decision-making, motor control.

• Information propagates through the brain.

• If it ignites the frontal cortex, it becomes conscious and available for action.

eg.Like a central server — only data uploaded to the server can be accessed by other systems.

Characteristics of Global Neuronal Workspace (GNW)

• Explains access consciousness (information available for use).

• Requires long-range connections (e.g., from visual cortex to frontal areas).

• Does not explain (hard prob) phenomenal consciousness (subjective experience).

Recurrent Processing Theory (RPT)

• Consciousness arises from recurrent (looping) activity in the brain. Looping can occur locally, regionally, or across the whole brain.

• Explains both phenomenal and access consciousness.

• Consciousness can occur without frontal activation.

• Attention and consciousness are separate processes.

eg.Playing tennis — you’re aware of the ball, racket, court, and surroundings due to multiple loops, even if you only attend to the ball.

Stages of Recurrent Processing Theory (RPT)

1. Feedforward Sweep: Visual signals are rapidly transmitted forward through the early visual system.Reflexive, unconscious processing.

2. Deep Feedforward Processing: Signals travel further up the visual hierarchy and can influence action. Not conscious. Enables motor responses and deeper processing. eg.Like data being processed in a pipeline — it’s moving but not yet accessible for reflection.

3. Superficial Recurrent Processing: info loops back to early visual areas (local feedback) — enables phenomenal consciousness (subjective experience of seeing).

4. Widespread Recurrent Processing: info loops through widespread areas, including frontal cortex — leads to access consciousness (percepts become available for decision-making, memory, and action—conscious manipulation of info).

Top-Down Attention

• Definition: Attention guided by goals, expectations, or prior knowledge. aka endogenous attention

• Example: Looking for your friend in a crowd.

Bottom-Up Attention

• Attention captured by salient or unexpected stimuli. aka stimulus-driven attention

• —>Features that drive salience: colour, shape, orientation (tilted line among vertical), motion 

• eg: Turning toward a loud bang. g.colour distractor slowed down responses to a shape target

⏱ Time Course of Attention

• Bottom-Up Attention: Fast and short-lived.

• Top-Down Attention: Slower but more sustained.

• eg— Fast eye movements were equally likely to go to target and distractor; slower movements favoured the target, showing bottom-up dominance early and top-down influence later.

Selection History

• Definition: Attention influenced by past experiences or learned associations, independent of goals or physical salience.

• Reward-associated stimuli act like bottom-up signals but are learned, so they don’t fit neatly into top-down or bottom-up categories

• Can apply to arbitrary stimuli through learning.

• Eg. A red strawberry captures attention because it’s associated with sweetness and reward. eg.Your attention is drawn to Doritos because you’ve often eaten them at rewarding social events.

Information Processing Bottleneck

• Definition: The idea that while sensory systems take in vast amounts of data, only a small portion is cognitively processed.

• Example: A pilot sees many dials in a cockpit but can only process one at a time.

Parallel vs. Serial Processing

• Parallel (Perception): Sensory systems process multiple inputs simultaneously.

• Serial (Cognition): Higher-level thinking processes one item at a time.

• Example: Seeing all letters on a screen but reading them one by one

-Attention connects perception (parallel) to cognition (serial), determining what gets processed.

Overt Attention

• Definition: Physically directing sensory organs (e.g., eyes, head) toward a stimulus.

• Example: Turning your head to look at a camera behind you.

Covert Attention

• Definition: Mentally focusing on a stimulus without moving your eyes or head.

• Example: Looking at a dot on a screen while mentally focusing on letters around it.

Link Between Overt and Covert Attention

• Covert attention often shifts to a location before overt attention (eye movement) does.

• Preparing to move your eyes to a target enhances perception at that location.

• before you move your eyes to a location (i.e., before you shift overt attention), covert attention moves to that location

Selective Attention

• Definition: Focusing on one stimulus while ignoring others.

• Example: Listening to a conversation in a noisy room

Change Blindness

• Definition: Failure to notice changes in a visual scene, even when they are large or obvious.

• eg.A person’s clothing color changes between frames.

• Key Conditions:

  • Blank screen sandwiches: A change is presented between two blank screens. (prevents Motion/Flicker Detectors -detect sudden changes or movement -from highlighting change)

  • Gradual changes: Slow transitions don’t trigger motion detectors.

  • Motion silencing: Movement masks changes in color, shape, or size.

Inattentional Blindness

• Definition: Missing an unexpected stimulus because attention is focused elsewhere.

• eg. Radiologists missing a gorilla embedded in a lung CT scan.

• Key Influences:

  • Attentional set: What you’re focused on affects what you notice.

  • Perceptual load: Harder tasks consume more attention, leaving less for unexpected stimuli.

🎯 Types of Attentional Selection

1. Location-Based Selection: Selecting stimuli based on their spatial location eg.Focusing on the left side of a screen -Posner Cueing — Valid cues improved performance; invalid cues reduced it

2. Feature-Based Selection: Selecting stimuli based on simple features like color or shape eg.triangles among squares.

3. Feature Combination Selection: Attempting to select stimuli based on a combination of features (e.g., red circles).

4. Feature + Location Selection -not possible!: Attempting to select stimuli based on both a feature and a location (e.g., red squares on the left). feature selection is global and cannot be restricted to a location. —Trying to attend only to red squares on the left results in attention to red squares everywhere.

Priority Map

• neural representation where top-down and bottom-up signals are combined to determine attentional focus.

• Top-Down Sources: Frontal and parietal areas.

• Bottom-Up Sources: Sensory areas like visual cortex

—> These signals are thought to converge in areas like the frontal eye fields, forming a priority map that ranks stimuli based on their combined relevance and salience.

Top-down and bottom-up signals can compete:

• A highly salient distractor (bottom-up) may override a goal-directed search (top-down).

What vs. Where Pathways

• Ventral Pathway (What): Object recognition; processes features.

• Dorsal Pathway (Where): Spatial location; processes positions.

• Implication: Difficulty in combining feature and location selection may arise from the separation of these pathways.

Object-Based Processing Limits

Processing multiple objects is harder than processing multiple features within one object.

• Participants made more errors when judgments were split across two objects (line and box) than when both judgments were about one object

Visual Search

• The process of scanning a visual environment to locate a specific target among distractors.

• Examples:

  • Finding your keys in a messy room.

  • Searching for a word in a block of text.

Types of Visual Search

• Parallel Search: A fast and efficient search where all items are processed simultaneously.

  • Search time does not increase with more distractors.

  • Occurs for simple features (e.g., color, shape).

  • Driven by bottom-up attention and pre-attentive processing.

• Serial Search: A slower, effortful search where attention moves item-by-item.

  • Search time increases with more distractors.

  • Required for feature combinations or complex shapes.\

  • eg. Finding Wally

Feature Combination

• Searching for a target defined by a combination of features (e.g., color + shape).

• Challenge: Cannot be selected directly; requires focused attention.

• Example: Searching for a red circle among red squares and blue circles.

Feature Integration Theory (Treisman & Gelade, 1980)

• Explains how features are processed and combined.

• Stages:

  • Pre-attentive stage: Simple features processed in parallel search.

  • Focused attention stage: To recognise objects made of multiple features, we need to focus attention on each item. Features are integrated into object representations. -This happens in serial search (feature search)

• Implication: Explains why conjunction searches are slower and more error-prone

Conjunction Search in Baggage Scans

• Searching for items defined by multiple features (e.g., shape + color + orientation).

• Challenges:

  • Low target salience.

  • Abstract target definitions.

  • Multiple target types.

  • Rare targets.

Ultra-Rare Item Effect

• When targets are very rare, people are more likely to miss them.

• Example: Baggage screeners miss weapons because they appear in <1% of bags.

• Finding: Miss rates increase as target prevalence decreases; response bias shifts toward “no target”.

Threat Image Projection

• A method to increase target prevalence by inserting fake threats into X-ray images.

• Purpose:

  • Prevent complacency.

  • Monitor screener performance.

• Effectiveness: Improves lab performance; real-world impact unclear.

Multiple Target Search

• Accuracy drops when searching for two targets simultaneously.

• Issues:

  • Hard to maintain two target templates.

  • Satisfaction of search: Finding one target reduces likelihood of finding another.

Search Performance Variability

• Some people are naturally better at visual search.

• Training:

  • Improves performance.

  • Does not eliminate individual differences.

• Example: Good searchers remain better even after extensive training.

Grand Illusion of Visual Experience

• We feel like we perceive the entire visual field, but we don’t.

• Argument:

  • Premise 1: If we experience most of a scene, we should notice changes.

  • Premise 2: We don’t notice changes.

  • Conclusion: We don’t experience most of the scene.

Attentional Set

• The mental template or category you’re focused on.

• Effect: Shapes what you notice.

• Example: Focusing on letters makes you more likely to notice an unexpected letter than a number.

Load Theory

• Definition: Attention has limited capacity, and all of it gets used.

• Low Load: Easy tasks leave spare attention for unexpected stimuli.

• High Load: Hard tasks use all attention, increasing inattentional blindness.

• eg.Harder basketball counting task led to lower gorilla detection

Residual Processing

• Definition: Even when we don’t consciously notice a stimulus, some processing may still occur.

• Example: Participants who said they didn’t see a red line still guessed its location above chance.

Value-Modulated Attentional Capture (VMAC)

• Definition: Stimuli paired with high-value rewards capture attention more than those paired with low-value rewards.

• Eg: A colored circle associated with a 10¢ reward captures attention more than one associated with 1¢.

• VMAC effect is strongest for fast eye movements, suggesting early-stage processing.

VMAC vs. Top-Down/Bottom-Up

• Not Top-Down: Participants know looking at the distractor is bad, but still do it.

• Not Bottom-Up: effect works even when distractors are equally physically salient.

• Conclusion: VMAC is driven by learned reward associations.

VMAC and Decision Making

• Attention influences decision-making by increasing the rate of evidence accumulation (drift rate).

• Example: If you attend more to Gatorade than orange juice, you’re more likely to choose Gatorade.

VMAC and Addiction 

• VMAC effects are linked to addictive and compulsive behaviours.

• Definition: Learned reward associations can lead to attentional biases toward substance-related stimuli.

• Examples:

  • Drinkers attend more to alcohol-related cues.

  • VMAC scores predict drug use, risky alcohol use, and Dry January success.

Cognitive Psychology

• Definition: The scientific study of mental processes such as perception, memory, attention, language, and problem-solving.

Information Processing Metaphor

• (1950s–1980s)

• Compares the mind to a computer that processes symbols (inputs) to produce outputs.

• Example: Remembering a phone number involves encoding it in short-term memory, rehearsing it, and transferring it to long-term memory—similar to how a computer processes and stores data.

Connectionist Framework / Neural Network Metaphor

• (1980s–Present)

• The mind operates like a network of interconnected neurons.\Mental processes emerge from patterns of activation and inhibition across simple processing units.

• Emphasizes learning through experience by adjusting the strength of connections (like synaptic plasticity).

Cognitive Neuroscience

• (1990s–Present)

• Understanding the mind requires studying the brain.

• Uses neuroimaging techniques (e.g., fMRI, PET, EEG) to link cognitive functions to specific brain regions.

• Emphasises localisation of function and brain-behavior relationships.

• but: The brain is like a city viewed from a satellite—you can see patterns of activity (like traffic) but not the internal workings of each building.

🧪 Approaches to Studying Cognition

1. Experimental Cognitive Psychology: Uses controlled experiments to infer mental processes from behavior (e.g., reaction times, error rates).

2. Cognitive Neuropsychology: Studies individuals with brain damage to understand normal cognitive functioning. (eg.HM’s case showed the hippocampus is critical for long-term memory but not short-term memory.)

3. Computational Cognitive Science: Builds computer models to simulate human cognitive processes.

4. Cognitive Neuroscience: Combines brain imaging with cognitive tasks to link brain activity to mental processes.

Seductive Allure of Neuroscience

• People find explanations more convincing when they include irrelevant neuroscience.

• eg adding “frontal lobe circuitry” to a bad explanation made it seem better to non-experts.

• 3D brain images increase perceived credibility of scientific claims.

• Even irrelevant neuroscience boosts perceived quality of explanations.

Sensory Memory and types

• Brief storage of sensory information.

• Types:

  • Iconic Memory: Visual sensory memory; lasts ~200 ms.

    • Example: people could recall a cued row of letters better than the whole grid, suggesting high capacity but rapid decay.

  • Echoic Memory: Auditory sensory memory; lasts a few seconds.

    • Example: Remembering the last few words someone said even if you weren’t paying attention.

Short-Term Memory

• Temporary storage system for small amounts of information (~7±2 items) for ~20 seconds.

• Code: Phonological (sound-based).

• Forgetting: Rapid, due to decay or interference.

Long-Term Memory and types

• Stores information indefinitely; capacity is unlimited.

• Code: Semantic (meaning-based).

• Types:

  • Semantic Memory: Facts and concepts (e.g., “Paris is the capital of France”).

  • Episodic Memory: Personal experiences (e.g., “My 10th birthday party”).

  • Procedural Memory: Skills and actions (e.g., riding a bike).

Multi-Store Model vs Traditional View of Short-Term and Long-Term Memory

Multi-Store Model proposes three distinct memory stores: sensory memory, short-term memory (STM), and long-term memory (LTM) (Attention, rehearsal, encoding). Assumes STM and LTM are separate and that STM is not influenced by meaning.

• Criticisms: Oversimplified; assumes linear flow. Doesn’t account for semantic/top-down effects.

The traditional view distinguishes STM and LTM based on capacity, rate of forgetting, and type of code.

supporting Evidence for the traditional view of short term memory and long term memory

• Capacity: Miller (1956) proposed the "magic number" 7 ± 2 items. Demonstrated in digit span tasks and letter recall.

• Decay: Brown-Peterson Task showed that without rehearsal, STM decays rapidly.

• Phonological Coding: Participants confuse similar-sounding letters (e.g., C, G, P) even when presented visually—suggesting phonological encoding.

• Semantic Coding in LTM: LTM stores information based on meaning, as shown in semantic memory tasks and recall of meaningful sentences.

Levels of Processing Theory

• Memory strength depends on depth of processing.

  • Shallow: Surface features (e.g., font, sound).

  • Deep: Semantic meaning, self-reference.

• Example: Remembering a word better when asked if it describes you vs. counting its letters.

Evidence Challenging the Multi-Store Model

A. STM is Sensitive to Semantic Information

• Evidence: STM is not purely a passive, phonological store; it is influenced by meaning.

  • When presented with items from the same semantic category (e.g., fruits), recall declines over trials due to proactive interference.

  • When the category changes (e.g., to professions), recall improves—indicating semantic processing in STM.

B. Levels of Processing Theory (Craik & Lockhart, 1972)

• Core Idea: Memory retention depends on the depth of processing, not the store it enters.

• Evidence:

  • Participants remember words better when they process them semantically (e.g., “Is this word meaningful?”) than when they process them shallowly (e.g., “Does it have 3 letters?”).

Baddeley’s Original Working Memory Model (1974)

Baddeley and Hitch proposed that STM is not a single passive store but an active system called Working Memory, composed of:

1. Central Executive:

   • Directs attention and coordinates other components.

   • Example: Switching between tasks or suppressing distractions.

2. Phonological Loop:

   • Stores and rehearses verbal/auditory information.

   • Subcomponents:

     • Phonological Store: Holds sound-based info briefly.

     • Articulatory Rehearsal Process: Repeats info to prevent decay.

   • Example: Repeating a phone number to remember it.

3. Visuospatial Sketchpad:

   • Stores and manipulates visual and spatial information.

   • Example: Mentally navigating a map or visualizing a shape.

Baddeley’s Working Memory Model Revised (2000)

added a fourth component:

4. Episodic Buffer:

   • Integrates information from the phonological loop, visuospatial sketchpad, and long-term memory.

   • Acts as a temporary storage system for multi-modal information.

   • Example: Remembering a story that includes visual scenes and spoken dialogue.

Executive Function (EF)

Executive functions are a set of cognitive processes that are responsible for controlling and regulating thought and action, especially in goal-directed behavior. These processes are closely associated with the frontal lobes of the brain and are essential for managing attention, inhibiting inappropriate responses, switching between tasks, and updating working memory contents.

Friedman and Miyake’s (2012) Alternate Approach to Investigating Working Memory

proposed an executive function (EF)-based approach to studying working memory, focusing on the control and regulation of thought and action rather than just storage capacity.

Key Components:

• Updating: Monitoring and revising working memory contents (e.g., replacing old items with new ones).

• Shifting: Switching between tasks or mental sets (e.g., alternating between classifying shapes and colors).

• Inhibition: Suppressing dominant or automatic responses (e.g., resisting distraction).

This approach highlights individual differences in executive control and links working memory to frontal lobe function, offering a more nuanced understanding than traditional capacity-based models.

Difference Between Simple and Complex Span Tasks

Simple Span Tasks:

• Measure storage only.

• Example: Digit Span – recall a sequence of numbers in order.

Complex Span Tasks:

• Measure storage + processing.

• Example: Operation Span – solve math problems while remembering unrelated words.

Why Complex Span Tasks Are More Useful

• Reflect real-world cognitive demands (e.g., reading comprehension, problem-solving).

• Higher correlation with cognitive abilities:

  • Daneman & Merikle (1996): Complex span tasks better predicted language comprehension than simple span tasks.

  • Complex math span tasks correlated more strongly with reading comprehension than simple verbal span tasks.

Working Memory Capacity and Its Relationship to Mind Wandering

• High-WMC individuals stayed on-task more during demanding activities.

• Low-WMC individuals reported more mind wandering, especially during tasks requiring effort and concentration.

Working Memory Capacity and Its Relationship to Distraction by Seductive Images

• Low-WMC participants were more distracted by seductive (irrelevant but interesting) images, leading to poorer comprehension.

• High-WMC participants were unaffected by image type.

Working Memory Capacity and Its Relationship to Focus and Attention

High WMC is associated with:

• Better attentional control.

• Greater resistance to distraction.

• Enhanced task engagement under cognitive load.

Cowan’s (2010) Argument: Working Memory Is Not Separate from Long-Term Memory + neuro evidence

• Working Memory (WM) is not a separate system but an activated subset of Long-Term Memory (LTM).

• Known as the Embedded-Processes Model.

Key Concepts:

• Focus of Attention: A limited-capacity spotlight (3–4 items) within activated LTM.

• WM = Activated LTM + Focus of Attention.

Neuroscientific Evidence for Cowan’s (2010) Argument: Working Memory Is Not Separate from Long-Term Memory + neuro evidence

Ranganath et al. (2003):

• fMRI study comparing WM and LTM tasks.

• Found overlapping activation in prefrontal cortex regions (e.g., BA 9/46, BA 10/46) during both WM and LTM tasks.

• Suggests shared neural substrates, challenging the idea of distinct

Declarative (Explicit) Memory

• Memory for facts and events that can be consciously recalled.

• eg.Recalling the capital of France

• measured using: 

  • Recall tasks (e.g., free recall, cued recall)

  • Recognition tasks (e.g., multiple choice, face recognition)

Procedural (Implicit) Memory

• Memory for skills and actions that are performed automatically, often without conscious awareness.

• eg.Riding a bike

• measured using

  • Performance-based tasks (e.g., mirror tracing, serial reaction time tasks)

  • Priming tasks (e.g., word-stem completion)

Difference Between Recall and Recognition & Effects of Context

Recall:

• Retrieving information without cues.

• eg.“What did you eat for lunch yesterday?”

• Effect of Context Change: Highly context-dependent. Performance drops if the retrieval context differs from the encoding context.

Recognition:

• Identifying previously learned information from options.

• eg.“Did you eat sushi or pasta for lunch yesterday?

• Effect of Context Change: Less context-dependent. Recognition is generally easier and more resilient to context changes.

Transfer-Appropriate Processing

idea that memory performance is best when the cognitive processes used during encoding match those used during retrieval.

Memory is not just about how deeply information is processed (as in levels of processing), but also about how appropriately it is processed for the retrieval task.

• Morris et al. (1977): Participants who encoded words based on rhyming (e.g., “Does ‘train’ rhyme with ‘brain’?”) performed better on a rhyming recognition test than those who encoded words semantically.

• Practical Example: If you study for a multiple-choice test by practicing multiple-choice questions, you’ll likely perform better than if you only used essay-style practice.

Evidence for Different Memory Systems from Amnesiacs’ Performance

Amnesiacs (e.g., H.M.) show severe impairments in explicit memory (e.g., free recall, recognition), but preserved performance on implicit memory tasks (e.g., word stem completion, mirror drawing).

Examples:

• Warrington & Weiskrantz (1970):

  • Amnesiacs performed poorly on explicit tasks but normally on implicit tasks like word fragment completion.

• H.M. (Henry Molaison):

  • Could not recall doing mirror drawing tasks but improved over time, indicating intact procedural memory.

• Fragment Completion Task:

  • Participants unknowingly complete word fragments with previously seen words, even if they don’t consciously remember seeing them.

Objections to the Different Memory Systems Explanation

A. Task Purity Problem

• Explicit memory tasks (e.g., recognition) are not process-pure—they involve both recollection and familiarity.

• Implicit memory may also be influenced by conscious processes.

• eg: Recognition responses include a mix of “remember,” “know,” and “guess” responses, showing overlap.

B. Transfer Appropriate Processing (TAP) Account

• Many dissociations can be explained by matching cognitive processes at encoding and retrieval, not by separate systems.

Examples:

• Generating a word (deep processing) improves explicit memory but impairs implicit memory because the word was never seen visually.

• Pictures helped explicit recall but not implicit word fragment completion—because the modality didn’t match.

difference between remember know evidence (recollection and familiarity) + implication for separate memory systems

• Remember: Conscious recollection of contextual details (episodic memory).

• Know: A vague sense of familiarity without specific details.

Key Study:

• Deep encoding (semantic) increased “remember” responses.

• “Know” responses were unaffected by encoding depth.

Neuroscience Evidence:

• Hippocampus activated only during “remember” responses.

• Parietal cortex activated during both “remember” and “know” responses.

• No brain areas were more active for “know” than “remember”.

Implication:

• Suggests recollection and familiarity are distinct processes, but not necessarily separate systems.

• “Remember” may require additional neural resources, while “know” is more perceptual/semantic.

Process Dissociation Method

To separate recollection and familiarity in memory tasks.

• Recollection estimate = Inclusion score − Exclusion score.

Neuroscientific Evidence Supporting the Memory Systems View

Neuroscience provides strong support for the idea that different memory types rely on different brain systems, particularly distinguishing recollection (explicit memory) from familiarity (implicit memory).

fMRI Studies:

• Recollection activates the hippocampus and medial temporal lobe (MTL).

• Familiarity activates parietal regions and shows less hippocampal involvement.

• Example: Multiple neuroimaging studies show many areas activated for recollection, but only one or few for familiarity.

Neuropsychological Data:

• Patients with MTL lesions show impaired recollection but preserved familiarity.

• Example: In recollection tasks, MTL-lesioned patients perform worse than non-MTL-lesioned patients; no difference is seen in familiarity tasks.

Pharmacological Evidence:

• Valium (a benzodiazepine) impairs recollection but not familiarity, suggesting different neural mechanisms.

⚔ Conflict Between Behavioural and Neuropsychological Evidence

⚔ The Conflict:

• Neuropsychological Evidence:

  • Supports distinct memory systems (e.g., hippocampus for recollection, other areas for familiarity).

  • Based on lesion studies and brain imaging.

• Behavioural Evidence (Healthy Participants):

  • Often supports processing-based accounts (e.g., transfer appropriate processing).

  • Suggests that task demands and encoding-retrieval match explain memory performance, not necessarily different systems.

🛠 Constructive Ways Forward from Conflict Between Behavioural and Neuropsychological Evidence

Integration of Views:

• Rather than viewing systems vs. processes as mutually exclusive, researchers propose a complementary approach.

• Example: McClelland’s Complementary Learning Systems:

  • Hippocampus: Rapid encoding of episodic memories.

  • Neocortex: Gradual learning of semantic knowledge (e.g., categories, generalizations).

Process Dissociation Methods:

• Separates recollection and familiarity within the same task.

• Helps clarify which processes are contributing to performance.

Computational Modeling (e.g., PDP models):

• Offers testable simulations of how different memory types might be represented and learned.

• Can simulate graceful degradation (e.g., semantic dementia) and category learning.

Hierarchical Network Models

• Structure: Concepts organized in a tree-like hierarchy (e.g., animal → bird → canary).

• meaning representation: Meaning is in the node (e.g., “canary” node contains all info about canaries).

• Strengths: Efficient for inheritance (e.g., canaries inherit properties of birds).

• Limitations: Doesn’t explain how nodes form or what’s inside them.

Associational Networks

• Structure: Concepts linked by associative strength (e.g., “dog” ↔ “bark”).

• meaning representation: Meaning is in the connections between nodes.

• Strengths: Explains priming and semantic activation.

• Limitations: Still assumes predefined nodes; lacks learning mechanism.

PDP (Parallel Distributed Processing) Models

• Structure: simulate how memory and cognition might emerge from networks of interconnected units (like neurons).

• meaning representation: Meaning is in the pattern of activation across the network.

• Strengths: - Explains generalization and graceful degradation.
  - Models semantic dementia and category learning.
  - No need for predefined nodes.

• Limitations: - Poor at episodic memory and one-trial learning.
  - Struggles with exceptions (e.g., penguins).

Principles of PDP (Neural) Networks: How They Learn 

• Distributed Representation: Concepts are not stored in single nodes but as patterns of activation across many units.

• Hebbian Learning: “Cells that fire together, wire together.” Co-activated units strengthen their connections.

• Error-Driven Learning:

  • The network compares its output to the correct answer.

  • Backpropagation adjusts the connection weights to reduce error.

  • Learning occurs gradually over many epochs (training cycles).

Principles of PDP (Neural) Networks: How They Perform 

• Generalization: The network can apply learned patterns to new, similar inputs (e.g., if it knows “canary” is a bird, it can infer that “sparrow” can fly).

• Graceful Degradation: If the network is damaged, performance declines gradually rather than failing completely—mirroring real-world conditions like semantic dementia.

Usefulness of PDP Models for Understanding Memory

✅ Generalization:

• PDP networks can infer new information based on similarity to known patterns.

• Example: Learning about “canary” helps the network understand “sparrow” without starting from scratch.

✅ Graceful Degradation:

• When damaged, PDP networks lose detail gradually, not all at once.

• Mirrors semantic dementia, where patients progressively lose the ability to name or distinguish between similar concepts (e.g., calling a “lion” a “tiger” or eventually just “animal”).

relate PDP models to the structure vs processes memory distinction debate

🔄 Structure vs Process Debate:

• In PDP models, structure and process are inseparable.

  • The “structure” (knowledge) is the pattern of activation.

  • The “process” is the activation and updating of those patterns.

• This contrasts with older models that separate memory into static structures (e.g., nodes) and dynamic processes (e.g., retrieval).

Limitations of PDP Networks

Embodied vs Abstract Representations

• Abstract Representations: Concepts are stored in amodal, symbolic forms, separate from sensory or motor systems (e.g., a “dog” is represented as a node in a semantic network).

• Embodied Representations: Concepts are grounded in sensorimotor systems—they are stored in the same brain regions involved in perceiving or interacting with them.

Evidence Supporting Embodiment

• Barsalou (2009): Argued that concepts like “bicycle” are not stored as single abstract entities but are context-dependent (e.g., riding vs. repairing vs. driving behind a bike).

• Hauk et al. (2004): Lexical decision tasks with words like “lick,” “pick,” and “kick” activated motor areas for tongue, finger, and foot, respectively.

• Pulvermüller et al. (2005): TMS stimulation of motor cortex areas facilitated responses to matching action words (e.g., arm-related words when arm motor strip was stimulated).