Cognitive Science Quiz 2 (Eliatamby) - Modules 4-6

Computation

Its inputs and outputs can be usefully and systematically interpreted as representing the ordered pairs of some function that interests us.

Pancomputationalism

All sufficiently complex physical systems perform computations

Formal System

A system that takes symbols, combines them into expressions, and manipulates them using processes

Computational Theory of Mind

The mind is a computer

What are Marr’s Levels of Analysis?

1. Computational

2. Algorithmic

3. Implementational

What does the computational level ask?

“What?” and “Why?”

Computational Level

Most abstract level; it looks at the goal of the computation, its inputs and outputs, and why the computation is well-suited for the task

What does the algorithmic level ask?

“How?”

Algorithmic Level

It asks how the inputs and outputs are represented, and looks at the specific steps (procedure) the process takes from input to output

What does the implementational level ask?

“Where?”

Implementational Level

It asks how the computation is realized in a physical system (hardware), what physical structures and processes are involved, and what’s the relevant level of description

Functionalism

Mental states are defined in terms of their causes and effects (alternatively, inputs and outputs)

Semantic properties (mental symbols have these)

Referring to or about things in the real world (they have meaning behind them)

Representation

Something that stands for something else

Features of a mental representation

Bearer, Content, Grounding, and Interpretability

Bearer

The thing that carries (makes) the representation

Content

What the representation says or means

Grounding

The relationship that connects the content to something in the real world

Referent

The actual thing in the world being represented

Interpretability

The fact that the representation can be understood or interpreted

What are the types of mental representations?

Concepts, Propositions, Mental Maps, Mental Images

Concepts

The building blocks of thought; concepts are symbolic (they stand for an idea or object but don’t have a genuine resemblance)

Proposition

Complex representations with sentence-like structure that can be true or false

What can propositions do?

Can represent logical relationships and counterfactuals, allows us to reason about the world, and help us think and talk about abstract things

Mental Map

Representation of spatial layout that captures info like direction and distance

Mental Image

1-1 correspondence; mental images are isomorphic

Aphantasia

An inability to create a voluntary visual mental image

Analog mental representation

Continuous and gradient

Digital mental representation

Discrete and categorical

Approximate Number System

Detect differences between large sets without counting

Subitizing System

Discriminate very small sets without counting (up to 4 in adults, 3 in children)

Weber’s Law

The discriminability of any 2 magnitudes is a function of their ratio

Successor Function

Every next number in a list is 1 more unit than the previous one

Turing Machine

A mathematical model that can compute anything; formalized algorithms and is the basis of all modern computing

Why was the Turing Machine created?

To prove the undecidability of the halting problem

Halting Problem

Given a program some input. Will the program finish running, or will it run forever? Can a general algorithm solve this problem for any program-input pair?

What are 2 defining traits of the Turing machine?

Automaticity, Determinacy

Automaticity

There’s no external operator calling the shots

Determinacy

Behavior is determined entirely by the current state and symbol

Church-Turing Thesis

Anything that is computable can be computed using a Turing machine

Multiple Realizability Thesis

1. Systems with minds are cognitive systems

2. Cognitive systems are computational systems

3. Turing machines can completely describe any computational system

4. Therefore, Turing machines can completely describe any cognitive system

5. Turing machines are defined independently of implementation

6. Therefore, cognitive systems can be defined independently of implementation

7. Therefore, systems with minds can be defined independently of implementation

Turing Test

Imagine you are communicating with a human and a computer through texts, both of which are hidden from you. You can ask them questions, but you can’t hear or see them, only their written responses. If you cannot reliably tell which is the computer, the computer is said to have passed the Turing Test.

The Frame Problem

How can we represent the effects of an action?
Problem: the set of possible effects and non-effects is immense
How can an AI list relevant effects without explicitly listing all non-effects?

Searle’s Room

A computer can follow rules to manipulate symbols and appear to understand a language without actually understanding it

The Complexity Problem

An algorithm designed for a physical system can only be run by another physical system with the addition of an emulator. The increase in computational complexity (via the emulator) can affect behavior.