Part 1: The Basics of Cognitive Psychology: Core Concepts

Welcome to Last Minute Lecture!

This free chapter overview is designed to help students review and understand key concepts.

These summaries supplement not replaced the original textbook and may not be redistributed or resold.

For complete coverage, always consult the official text.

Okay, let's unpack this.

I want you to try something for me.

Right now, wherever you are,

just look at the nearest object.

Maybe it's a coffee mug, a phone, or I don't know, just your own hand.

You look at it and you know what it is instantly.

Right.

It feels seamless.

It feels automatic.

It feels simple.

But if you actually stop to think about what just happened inside your skull to make that happen, it's staggering, is a miracle of engineering that makes the most advanced supercomputer look like a pocket calculator.

It really is.

It's the ultimate black box.

We all have this incredible machine, the human mind, but we rarely get a look at the wiring diagram.

We just assume it works.

We take the user interface of our consciousness for granted.

Exactly.

And that is our mission today.

We're cracking open that black box.

We're doing a deep dive into part one of Carol Brown's cognitive psychology.

And I want to be clear about what this is.

This isn't just like a random collection of facts.

This text serves as a starter kit for the entire field.

If you are a student encountering cognitive psych for the first time, or just someone who wants to know how their brain actually processes the world,

this is the roadmap.

And it's a vital roadmap because it sets the stage for everything that comes later.

You know, perception, memory, language, decision making.

All of it.

None of those chapters make sense unless you understand the foundation laid out right here in part one.

We're going to how to think like a scientist to make sense of the mess inside our heads.

So let's start with the central metaphor.

The source material makes a really strong defining comparison right out of the gate.

It does.

It says if you want to understand the mind, you can't just look at the squishy biology.

You have to look at it like a computer.

Right.

This is the information processing approach.

It is the dominant paradigm that the text describes.

Just like your laptop, the human mind is seen through three fundamental stages.

You have input, which is all the information flooding in through your eyes, your ears.

Then you have storage, which is the whole complex system of memory.

And finally,

retrieval.

That's the output, the decision you make based on all that data.

Input, storage, retrieval.

That seems almost too clean, doesn't it?

I mean, human beings are messy.

We have emotions.

We have irrational thoughts.

Can we really be boiled down to a data processing flow chart?

That's a fair pushback, and it's one that critics have raised for decades.

But the text contrasts this computer approach with the older way of doing things.

Which was called introspection.

Exactly.

And to understand why the computer metaphor won, you have to understand why introspection just failed.

Introspection.

That's just looking inward, right?

Like, how do I feel about this?

Essentially, yes.

It was the early method where researchers would train people to describe their own mental processes, you know, describe the experience of seeing the color red.

Now, the text notes this is useful for description.

If you tell me a sunset makes you feel melancholy, okay, I believe you.

But for scientific interpretation, It's useless.

It's completely broken.

Why?

Because we lie to ourselves?

Or because we're just unreliable narrators?

It's even deeper than that.

It's that we are usually completely unaware of our own mental processes.

When you recognized that coffee cup earlier, did you feel your retina detecting the edges?

Did you feel your memory searching for the concept of a handle?

No, of course not.

I just saw the cup.

Exactly.

The processing is hidden from you.

You only get the final result.

If we relied on people telling us how they think, we'd get a mic of the world drawn by someone who's never left their basement.

So cognitive psychology assumes we are like computers to move the study into the realm of the empirical.

We use controlled conditions to infer what's happening in the software based on the behavior we can actually measure.

Okay, that makes sense.

We treat the mind as an information processor because it gives us something we can actually test.

But this field doesn't exist in a vacuum.

Not at all.

The text describes this massive ecosystem of related disciplines that all feed into cognitive psychology.

It's like a family tree, and some of the relatives are pretty heavy hitters.

It is convergence of fields.

And understanding that ecosystem is vital, you have cognitive science, which is the heavy hitter on the computer models.

The text outlines three specific ways these scientists model the mind to predict behavior.

And these aren't just abstract ideas, they're like blueprints for how we might be wired.

Let's break those down because this is where the text gets a bit technical.

The first one it mentions is semantic networks.

Okay, so imagine a massive spider web.

In a semantic network, concepts are nodes.

So picture a dot representing bird, another dot representing wings, and another for Robin.

And the strands of the web are the connections.

They're the links.

So knowledge is basically geometry.

It's just how close things are to each other.

In a way, yes.

The links represent relationships.

Has wings, can fly, is yellow.

And the key here is learning is the act of adding new strands or changing the weight of the connection.

Give me an example of that weight changing.

Okay.

Let's say you're a child and you learn that a penguin is a bird,

but you also know birds fly.

Suddenly you have a conflict.

You have to adjust the network.

You create a link between penguin and bird, but you have to weaken or even sever the link between penguin and fly.

The whole web shifts.

So it's a very fluid way of mapping knowledge.

Okay.

That feels intuitive.

You build a web of ideas.

Then there's the second model,

production systems.

This one sounds a bit more bureaucratic.

It is much more rigid.

Think of it as a rule book.

It operates on if -then statements.

If this, then that.

Pretty much.

And it involves two types of memory working together.

You have long -term memory, which holds all the rules, like if the light is red, stop the car.

And then you have working memory, which holds the current situation.

The light I'm looking at right now is red.

So the system acts like a matchmaker.

It looks at the current situation, scans the rule book, finds the right match, and then executes the can.

Correct.

It matches the if condition to the then action.

It's a very logical,

algorithmic way of viewing behavior.

It explains those automatic habits we all have.

If coffee smell, then walk a titian.

You don't decide every single step.

The rule just fires.

Right.

And the third one, this one sounded the most complex and honestly the most relevant to what's happening in tech right now, connectionist networks.

Yes, often called neural networks or PDP parallel distributed processing.

This model abandons the rule book idea and tries to mimic the physical brain.

Imagine layers of nodes connected a bit like neurons.

The input comes in, say, an image of a cat, and it triggers a pattern of activity across these layers to produce an output.

That's a cat.

That's a cat.

But the text mentions a specific term here that I see thrown around in AI headlines all the time, backpropagation.

You hear people say this is how chat GPT learns or how Google Translate works, but the text treats it as a psychological concept.

What is it really?

Backpropagation is the secret sauce of learning in these networks.

And it's a tough concept, so let's use an analogy.

Think of it like a choir, and you are the conductor.

Okay, I'm on the podium.

Betten in hand.

You want a specific sound, a perfect C major chord, that's your target output.

You wave the baton, that's the input, the choir things, but it sounds awful.

The tenors are flat, the altos are screaming, and the basses are late.

The sound you actually get, the actual output,

is a mess.

So the error is the difference between that messy sound and the perfect chord I wanted.

Exactly.

Now here is backpropagation.

You don't just yell, do better, at the whole room.

You send a specific signal backwards through the choir, you point to the tenors and say, pitch up by 10%, you look at the altos and say volume down by 20%, you adjust the weights of the individual contributions.

If I do that a thousand times.

Eventually, the input produces the perfect output.

That is how a neural network learns.

It compares its guess to the correct answer, it calculates the error, and then it propagates that error backward to adjust the connections.

Wow.

It's entirely mathematical, yet it mimics how we perfect a tenacer or learn a language.

It's just trial, error, and micro -adjustment.

That makes it click.

It's not magic, it's just relentless calibration.

So that's the computer modeling side cognitive science.

But then we have the broken machine approach, which is cognitive neuropsychology.

This is a classic method.

The logic is, if you want to understand how a complex machine works, look at what happens when a specific part breaks.

By studying patients with brain damage, we can infer the function of healthy brains.

And the tech spends a significant amount of time here on the concept of dissociations.

Specifically, the difference between single dissociation and double dissociation.

This felt like one of those things that students probably get wrong on the exam.

It is the most common stumbling block, but it's the most critical tool in the detective kit.

Let's play this out.

Let's say you're a researcher.

You meet patient A.

Patient A has damage to a specific part of the brain.

We'll call it area X.

Now, patient A cannot recognize faces.

They look at their spouse and see a stranger.

That's terrible.

It is.

But they can recognize objects, like a lamp or a car.

Okay.

It seems open and shut.

Area X is the face recognition box.

It broke, so now they can't see faces.

That's the intuitive leap.

But a skeptic, and scientists are professional skeptics, would say, hang on, maybe recognizing a face is just harder than recognizing a lamp.

I see.

So maybe area X isn't a face center at all.

Maybe the whole brain is just weak because of the damage, and it fails at the difficult stuff first.

Exactly.

Like a car struggling to go uphill.

It's called a resource artifact.

Maybe patient A is just generally running on low battery.

To prove that faces and objects are processed by totally different systems, you need the Holy Grail.

You need a double dissociation.

You need the mirror image.

You need patient B.

Patient B has damage to a different area, area Y.

Patient B is great at faces.

They can pick their mother out of a crowd instantly.

But show them a lamp.

And they have no idea what it is.

No idea.

That rules out the difficulty argument.

Because if objects were just easy, patient B should get them right.

But they don't.

Exactly.

By finding these two opposite patients, you prove that the brain has distinct independent postal codes for faces and objects.

One can break while the other stays intact.

That is double dissociation.

It's the strongest evidence we can get without, you know, cutting open a healthy brain.

However, the expert in you has to point out the caveat the text offers.

We can't just map the brain perfectly this way, can we?

No, not at all.

The text warns us, modules don't strictly exist as little isolated physical boxes in the brain.

The brain is plastic.

It changes.

And relying on case studies of individuals makes it very hard to the whole population just because everyone's brain damage is unique.

It's messy.

Speaking of the physical brain, we move on to the third branch of the ecosystem,

cognitive neuroscience,

the hardware.

The text lists a few tools, single unit recording, EEG, PT, MRI.

These are the windows into the black box.

Single unit recording is incredibly sensitive.

It looks at individual neurons firing.

It gives you amazing detail on where and when, but it's usually very invasive.

And then EEG, that's the swimming cap with all the wires.

Yes.

EEG records electrical activity on the scalp.

It is fantastic for knowing when something happens.

It has great temporal resolution.

But it's terrible at telling you where it happened deep in the brain.

It's like listening to a party from outside the house.

You know, when the beat drops, we don't know who dropped it.

So then we use PT and MRI for the where.

Exactly.

Pete uses radioactive tracers in the blood.

MRI uses magnetic fields to track oxygen.

They give us those beautiful 3D images of lit up brains.

But the text reminds us to be cautious.

Very cautious.

These are indirect measures.

They show us blood flow, not thoughts.

And practically, you have to lie perfectly still in a giant noisy metal tube.

That is not exactly a natural environment for thinking.

Right.

Okay.

So that's the landscape, the ecosystem of cognitive science, neuropsychology, and neuroscience.

Now I want to hop into the time machine.

Section two of our deep dive is the history.

And according to the source, there was a very specific big bang moment for cognitive psychology.

The year is 1956 and 1957.

Why then?

What was happening before that?

Well, before this, American psychology was just dominated by behaviorism.

B .F.

Skinner.

Think of B .F.

Skinner.

The idea was that the mind doesn't matter.

You input a stimulus, you get a response.

What happens inside that black box?

Irrelevant.

But in 56 and 57, you had a rebellion.

A convergence of brilliance at a symposium at MIT.

Who were the rebels?

It was an all -star lineup.

You had Noam Chomsky, who basically just destroyed the behaviorist view of language.

He showed that language is way too complex to be just learned habits.

We must have an innate mental structure for it.

And George Miller.

George Miller published one of the most famous papers in history.

The magical number seven, plus or minus two.

Which is the idea that our short -term memory is limited.

Exactly.

We can hold about seven chunks of information.

This was revolutionary because it quantified a mental limit.

It treated memory like a system with specific specs.

Then you had Newell and Simon presenting the first true AI program, the logic theorist.

It was the moment the information processing metaphor became real.

So the focus shifted from what is the animal doing to how is the animal processing this?

Exactly.

And this solidified in the 1970s, which the text calls the consolidation period.

This is when the information processing paradigm really took the throne.

And what were the assumptions then?

The key assumptions became fixed.

People interact purposefully with the world.

We use symbols that have meaning.

And crucially, our processing takes time and has limited capacity.

We aren't infinite.

We have bandwidth limits.

Which is a perfect segue to section three, the hall of fame.

The source material outlines the specific models that came out of this revolution.

These are the models that every student absolutely needs to know.

Let's start with memory.

You have to start with Atkinson and Schiffer in 1968, the multi -store model.

I always picture this as a factory conveyor belt.

That's a good visualization.

It's a linear flow.

Information hits sensory memory first.

That's extremely brief, just a snapshot of sight or sound.

Then if you pay attention, it moves to short -term memory.

But here's the catch.

To keep it there, you have to rehearse it.

Like repeating a phone number over and over until you dial it.

555 -0199, 555 -0199.

Exactly.

That is the maintenance rehearsal loop.

If you rehearse it enough or process it deeply, it moves to the final box.

Long -term memory.

Simple, elegant.

Simple, elegant.

But as we discussed with the intro, simple isn't always right.

It was a bit too passive.

Enter Batley and Hitch in 1974.

They looked at that middle box short -term memory and said, no, it's not just a waiting room.

Right.

They proposed the working memory model.

They argued that short -term memory is an active workspace.

It's where we manipulate information.

And they broke it down into components.

OK, so we have the central executive at the top.

The boss.

The central executive is modality -free.

It doesn't care if the info is sound or sight.

And it controls the limited capacity of the system.

It directs traffic.

It says, focus on this math problem.

Ignore the dog barking.

And it has two slave systems working for it.

The names here are great.

The phonological loop.

Think of that as your inner voice and your inner ear.

It handles speech -based storage.

When you read a book and you hear the words in your head, that's the loop spinning.

And the visual -spatial scratch pad.

The inner eye.

Let's try another experiment.

Close your eyes.

I want you to visualize the front door of your childhood home.

OK.

Now visualize walking through it and turning left.

What do you see?

I see the staircase.

That image.

You are generating that on your visual -spatial scratch pad.

You are actively manipulating spatial information in your mind.

Battley and Hitch showed us that memory isn't just storage.

It's a dynamic workbench.

So that's memory.

Let's move to attention.

We have a battle of the titans here.

Broadbent versus Triesman.

Broadbent came first in 1958 with filter theory.

He proposed the bottleneck.

How does the bottleneck work?

Imagine two voices talking to you at once.

They both enter a sensory buffer.

But Broadbent argued there is a strict filter early in the process based on physical characteristics like pitch or loudness.

One voice gets through the filter to be processed for meaning.

The other blocked,

rejected.

It hits a wall.

I have to push back on that, though.

Please do.

Because if that were true,

cocktail parties would be impossible.

Or at least very boring.

Right.

If I'm talking to you and I'm 100 % focused on your voice, Broadbent says I literally shouldn't hear anything else.

But if someone across the room shouts fire or even just says my name.

Your head snaps up?

My head snaps up.

You just identified the exact flaw that broke Broadbent's theory.

If the filter was total, you would never hear your own name in a crowd.

You'd be a robot.

So Broadbent was just wrong.

He wasn't wrong about the bottleneck, but he was wrong about the seal.

Enter Anne Triesman, 1960.

She looked at that cocktail party effect and said, it's not a wall.

It's a volume knob.

Her attenuation theory.

Yes, attenuation just means turning it down.

She argued that the unattended signal isn't blocked.

It's just attenuated.

Turn down.

If a signal is weak but highly meaningful, like your own name or words that fit the context, it can still punch through the filter and trigger recognition.

So the threshold for my own name is really low.

Very low.

So even a quiet signal triggers it.

That makes so much more sense.

We don't have eyelids.

We just have volume control.

Okay.

Last stop in the Hall of Fame.

Perception.

How do we actually see the world?

The text sets up a massive debate between bottom -up and top -down processing.

This is fundamental.

On one side, you have James Gibson.

He's the champion of bottom -up or direct perception.

So data -driven.

Completely.

Gibson argued that we don't need to process or guess much.

The environment provides all the information we need.

The light hitting our eyes.

The optic array contains rich data.

He talks about optic flow.

That's the Star Wars effect, right?

When the stars streak past the windshield.

Exactly.

The speed and direction of things moving past you tell you exactly how fast you're going and where you are headed.

You don't need to calculate it.

You just see it.

He also coined the term affordances.

I love this concept.

It's the idea that when you see an object, you inherently see its function.

You don't see a flat raised wooden surface.

You see a place to sit.

The chair affords sitting.

It's direct.

But then you have Richard Gregory.

And he says, hold on, Gibson.

The world is messy.

Gregory takes the top down view.

He says sensory data is often incomplete or ambiguous.

We have to construct reality.

We use schemas, our past experiences and expectations to fill in the blanks.

This explains optical illusions.

Exactly.

Gregory loves illusions.

The reason you can look at a 2D drawing of a cube and see a 3D box is because your brain is actively forcing a 3D model onto the data.

You're guessing.

You are guessing.

Usually you guess right.

With illusions, you guess wrong.

And then there's Marr who seems to try to bridge the gap with a computational approach.

David Marr is the peacemaker in a way.

He breaks vision down into a hierarchy.

First, the primal sketch, just dealing with light, shadow and edges.

Then the 2 .5D sketch.

2 .5D.

Why not 3D?

Because it's centered on you.

The 2 .5D sketch maps depth and orientation relative to the viewer.

It's your perspective.

Oh, OK.

Finally, the brain constructs the full 3D model, which is the object's true structure independent of where you are standing.

It's a step -by -step assembly of reality.

It's amazing how much is happening just when I look at a coffee mug.

OK, we've covered the models.

But this text isn't just a history book.

It's a how -to manual.

Section 4 is all about thinking like a cognitive psychologist.

This is arguably the most practical part of the source material.

If you want to study the mind, you have to respect the scientific method.

And that starts with the hypothesis.

OK, I want to pass this course.

I need to think like a scientist.

Let's build an experiment right now.

I'm ready.

What's our hypothesis?

I have a theory.

I think that listening to heavy metal music helps you memorize poetry.

OK,

bold.

Heavy metal improves poetic retention.

That is a precise, testable statement.

Now, what are your variables?

The music is the thing I'm changing.

The independent variable, the obvious.

The cause.

You manipulate it.

Right, so I'll have one group listen to Metallica and one group listen to silence.

You could use silence, but then you have a problem.

If the silence group does better, is it because of the silence or just because they weren't distracted?

To really test the genre of heavy metal, you might want a control group listening to classical music.

Good point.

OK, Metallica versus Mozart.

And then I measure how many lines of Keats they can remember.

That is your dependent variable.

The DV, the effect, the data you collect.

Now, where do we do this?

In a soundproof booth.

Our lab seems safest.

No distractions.

It gives you control.

That is a lab experiment.

But does listening to Metallica in a sterile white room with a guy in a lab coat watching you tell you anything about how kids actually study in their bedroom?

Not really.

It feels fake.

That's the ecological validity problem the text keeps hammering on about.

A field experiment going into a dorm room and swapping their playlists would have higher ecological validity, but you lose control.

Maybe their roommate walks in.

Maybe they fall asleep.

It's always a trade -off.

Control versus realism.

Now, what if we can't run an experiment?

What if we just want to observe?

That's where non -experimental methods come in.

We talked about case studies like patient A.

You also have questionnaires for attitudes.

And you have correlation.

The text warns about correlation, doesn't it?

It does.

It screams it.

Correlation measures the relationship strength between two variables.

But, and this is the golden rule of psychology,

correlation does not mean cause and effect.

Give me the classic example.

Ice cream sales and shark attacks.

They are highly correlated.

When ice cream sales go up, shark attacks go up.

So, eating ice cream makes you tasty to sharks?

No.

There is a third variable, summer.

When it's hot, people buy ice cream and people swim in the ocean.

The variables move together, but one does not cause the other.

If you say correlation implies causation on your exam, you will fail.

Noted.

And briefly, we can't talk about methods without mentioning ethics.

Absolutely.

The text is clear.

Informed consent.

No deception unless absolutely necessary, and even then, you must debrief.

The protection of participants from harm is paramount.

You can't just break a mind to see how it works.

Which brings us to the final hurdle.

Assessment.

How does a student actually pass this course?

The expert in you must love this part.

I do, because it reveals the trap that so many students fall into.

The text is very explicit about the grading criteria.

To get a high score, a first class degree, 70 plus marks, you cannot just describe.

Yeah.

He said she said.

Exactly.

You can't just list what Battley said and then what Broadbent said.

That gets you a 40, maybe a 50.

The goal is critical evaluation.

What does that look like in practice?

It looks like saying Broadbent's theory was groundbreaking, however, it failed to account for the cocktail party effect, which suggests Treisman's attenuation model is more robust.

It means comparing, contrasting, and explaining why a study supports a theory.

So 40 to 49 marks is just basic understanding.

70 plus is original thought.

Original thought and wider reading.

You need to show you can build an argument, not just repeat a textbook.

To help with that, the text provides a cheat sheet of running themes.

These are concepts that pop up in every single chapter.

If you can connect these dots, you're winning.

Right.

We've touched on some bottom up versus top down processing is everywhere.

In memory, in perception, even in reading.

Modularity, the idea of independent systems like faces versus objects.

Chunking and rehearsal for learning.

And ecological validity.

That's the constant question.

Does this lab test actually apply to real life?

If you ask that question in your essays, you are thinking like a cognitive psychologist.

So we've covered a lot of ground.

We've defined the mind as a computer, an information processor, using input, storage, and retrieval.

We've looked at the ecosystem of science that supports it.

We've met the Titans, Chomsky, Miller, Atkinson, Schifrin.

We've looked at the tools from MRIs to semantic networks.

And we've learned that the secret to an A is critical evaluation.

It's a comprehensive starter kit.

It frames everything else you will learn in the field.

But before we go, I have one thought that's been nagging me.

We started with this metaphor of the production system, if X, then Y.

A rule -based machine.

Yeah.

If our minds are just networks of nodes and production rules.

If we are just processing inputs to create predictable outputs.

How much of our original thought is actually original?

Are we just running a really complex algorithm and calling it free will?

That is the ultimate question.

If we can map the network perfectly, do we become predictable?

Does the ghost in the machine disappear?

The text doesn't answer that.

It gives us the tools to ask it.

And that is what makes this field so exciting.

And that is a perfect place to leave it.

Something to chew on while you process your next caffeine input.

Thank you so much for listening to this deep dive into the basics of cognitive psychology.

It's been a pleasure.

From the whole last minute lecture team, thanks for being here.

Catch you on the next deep dive.