PSYC236 6.2

MODELS OF COGNITION
INFORMATION PROCESSING AND CONNECTIONISM

  • Course Information: PSYC 236: Thinking & Seeing

  • Instructor: Simone Favelle

Learning Outcomes

  • You will learn to:

    • Talk about and compare two ways our brain thinks: the "Information Processing" way and the "Connectionism" way.

    • Give an example of each way our brain thinks.

    • Say what's good and what's tricky about these two ways our brain thinks.

  • Reading: Eysenck & Keane (2015, 2020), Chapter 1.

Information Processing Approach

  • This idea says our brain is like a very smart computer or calculator (Marr, 1982).

  • It means thinking is like solving puzzles or following steps, guided by a set of rules.

  • Important Note: You can't understand just one part of your brain or thinking; all the parts work together like a team.

Representations and Processes

  • Two main things in our "brain computer":

    • Representation: This is like a picture or an idea you have in your head that shows something from the real world or a part of it.

    • Processes: These are the active steps your brain takes to change those pictures or ideas into new ones.

Processing Visual Information

  • When you see something, your brain does these steps:

    • Retinal Image: Your eye first sees a picture.

    • Image-based Processing: Your brain starts figuring out what the picture looks like.

    • Surface-based Processing: Your brain figures out what the surfaces of things are.

    • Object-based Processing: Your brain figures out what the whole object is.

    • Category-based Processing: Your brain knows what kind of thing it is (like, "it's a dog!").

  • This idea comes from Marr (1982).

Information Processing Model of Object Recognition

  • This is like a step-by-step plan for how your brain recognizes things (Riddoch & Humphreys, 2001).

  • Parts of this plan:

    • Structure: This is how the ideas and brain actions are connected together in an organized way.

    • Evidence Base: Scientists learned this from a book by Riddoch, M., & Humphreys, G. (2001) called Object Recognition.

Image Structural Descriptions and Semantic Knowledge

  • Your brain uses these parts to know objects:

    • Names: To know what to call things.

    • Descriptions of Objects: All the detailed information about the things.

    • Early Visual Processing: The first methods your eyes and brain use to see something.

    • Mapping Procedures: The techniques your brain uses to connect one idea or picture to another.

  • This idea comes from Riddoch & Humphreys (2001).

Neurophysiological Evidence

  • Feature Analysis: Studies where scientists looked at cells in the part of the brain that sees (called the visual cortex) have shown how our brain works to see objects.

    • Reference Study: A famous study by Hubel & Wiesel (1962) where they watched brain cells in cats to see how they worked.

Stimulus and Response Analysis

  • Stimulus-Response Series: This shows how different things (like a line you see) make brain cells react.

    • Tuning Curve Representation: This shows how much a brain cell "lights up" or gets excited when it sees a line pointing in different ways.

    • Visual Example:

    • Imagine a cell that loves lines. If the line is pointing straight up, the cell gets very excited. If the line is pointing sideways, it might not get excited at all.

Clinical Evidence: Visual Agnosia

  • Definition: Agnosia means someone can see with their eyes (their eyesight is fine), but their brain can't tell them what they are looking at or recognize objects visually.

  • Characteristics:

    • The problem is with the brain's understanding and thinking about what it sees, not with the eyes themselves.

    • People with Agnosia are still smart; their brain just has a special puzzle it can't solve when it comes to seeing and recognizing things.

Case Study: Sacks (1970)

  • Case Example: A famous story about "The man who mistook his wife for a hat."

    • This story shows how hard it was for him to know what simple things were, even when he could see them with his eyes.

Different Types of Agnosia

  • People with Agnosia can have different kinds of problems with recognition:

    • Apperceptive Agnosia: They can't tell what an object is based on its basic geometric features, like knowing a circle is different from a square.

    • Associative Agnosia: They can perfectly see all the features of an object, but their brain doesn't tell them what the object is for or what it means.

Detailed Features of Apperceptive Agnosia

  • Patients CANNOT:

    • Recognize objects.

    • Copy simple shapes, even if they are right in front of them.

    • Match two shapes that are the same.

  • Patients CAN:

    • Name colors.

    • Navigate through space (walk around without bumping into things).

    • Distinguish brighter and darker areas.

    • Detect the edges of shapes.

  • Identification Process: To try and figure out what objects are, these patients use other hints like color, how big something is, its texture, or if it's shiny, because they can't understand the basic shape differences.

Further Case Study: Herpes Simplex Encephalitis Patients

  • Observation: These patients found it hard to recognize or describe mostly natural or living objects (like animals or plants).

    • Case Study: Giulietta (Sartori et al, 1993).

    • She could tell the difference between overlapping drawings but couldn't put parts together to remember a whole object from memory.

    • Limitations: She had trouble saying what she saw in words, but she could still make general decisions about things.

Associative Agnosia

  • Definition: Patients can see perfectly fine, but their brain doesn't tell them what the thing means or what it's used for (known as “perception without meaning”).

  • Evidence: Stories about patients who could match and copy objects successfully but couldn't say their names.

    • Example Patient - JB:

    • He could pick out real objects from pictures, which means his brain could still see and know things, but he had trouble saying their names.

  • Task Examples:

    • If asked to pick out a real object, he could do it, even if he couldn't name it. This shows he still knew what real objects looked like.

    • But when asked to match objects based on their use or meaning (like matching a spoon to a fork), he struggled, even though his language knowledge was fine. This points to a problem with visually understanding what things are for.

Anomia

  • Definition: Patients with Anomia can see things, know what they are, match them, copy them, and describe them. Their visual and general knowledge is intact.

  • Key Deficit: Their big problem is that they just can't remember the names of things, people, or places. This makes them often use general words like "thingy" or "stuff."

Summary of Information Processing Models

  • Representation and Processing Summary:

    • Information processing models help us understand the steps and ideas our brain uses to recognize objects.

    • What we've learned from brain studies (neuropsychology) supports the idea that our brain recognizes things in structured stages. This makes us wonder about the limits of these models and look into other ways our brain might work.

The Connectionist Model

  • Definition: This idea says our brain is like a giant network made of billions of tiny connected brain cells (called "Neural Networks"). All our thinking comes from these connections.

  • Biological Plausibility: This idea makes more sense for how fast our brain thinks. If every thought had to go through many separate steps, thinking would take too long.

Connectionist Models and Brain Structure

  • Parts of this brain model:

    • Units: These are like tiny individual brain cells (neurons).

    • Activation: This is like how busy or excited these brain cells are, or how fast they are firing.

    • Connections: These are like the roads or links between the brain cells (synapses).

    • Connection Weight: This tells us how strong or weak these connections are. A strong connection might make another cell more excited, while a weak one might make it less excited.

Parallel Distributed Processing (PDP)

  • How processing works:

    • A huge number of small, simple helpers (units) do their simple jobs all together and at the same time.

    • Everything happens at once (in parallel), so different parts of a problem can be solved simultaneously, or at the same time as the whole input is being looked at.

Representation Coding Types

  • How ideas are stored:

    • Ideas can be stored in single units (like one helper for one idea) or as spread-out patterns of activity among many helpers.

  • Localized vs. Distributed Representation:

    • Localized Representation: An individual helper (node) means one specific thing (e.g., one node for "dog," one node for "bark").

    • Distributed Representation: The idea is stored across many helpers, with some of them holding small parts of the idea and others possibly not having a clear meaning on their own.

Advantages of Distributed Representations

  • Economy: It can store a lot of information using fewer helpers.

  • Generalization: It helps the brain understand new things better by using what it already knows (both general and specific information).

  • Learning Dynamics: It explains how our grown-up brain learns and changes over time.

  • Graceful Degradation: If some helpers get hurt, the whole brain doesn't stop working. It just works a little less well, depending on how many helpers are hurt.

Disadvantages of Distributed Representations

  • Interpretation Difficulty: It's hard to figure out exactly how all the helpers are working and what they are doing inside the network.

  • Superficial Brain Resemblance: This model might not be very similar to how real brain cells work because real brains have many different types of cells.

  • Scope Limitations: It doesn't fully explain all the complicated things our brain does, like feelings or how we get along with other people, very well.