Mind Design: Philosophy, Psychology, and Artificial Intelligence

Flashcards covering key vocabulary and concepts from the lecture 'What Is Mind Design?' by John Haugeland. Topics include definitions of mind design, artificial intelligence, Turing test, intentionality, formal systems, computers, algorithms, heuristics, GOFAI, connectionism, embodied/embedded AI, and the concept of understanding.

Mind Design

The endeavor to understand mind (thinking, intellect) in terms of its design (how it is built, how it works), often involving building something and making it work, rather than just observing what exists.

Artificial Intelligence (AI)

The attempt to construct intelligent artifacts or systems with minds of their own, considered central to mind design.

Mind Design as Reverse Engineering

A perspective where understanding the mind is approached by attempting to build or reconstruct it, similar to disassembling a product to understand its function.

Materialists (in Mind Design)

Those who suppose that matter, suitably selected and arranged, is sufficient for intelligence, believing intelligence does not depend on anything immaterial or supernatural.

Turing Test

A pragmatic criterion proposed by A.M. Turing where a nonhuman system is deemed intelligent if it can act so like an ordinary person in certain respects (specifically, via verbal behavior) that other ordinary people cannot distinguish it from a human.

Intentionality (Brentano's definition)

The 'mark of the mental,' referring to the character of one thing being 'of' or 'about' something else, such as by representing, describing, or aiming at it.

Intentional Object

That which an intentional act or state is 'about' or 'stretches toward,' which may or may not actually exist.

Dennett's Materialist View of Intentionality

The idea that intelligence and intentionality are a matter of how a system behaves, understood through holism (behavior in context) and interpretation (construing consistent behavioral patterns as rational).

Rationality (Dennett)

Acting in a way that best satisfies one's goals overall, given what one knows and can discern about the situation.

Intentional Stance

The perspective adopted when regarding a system as having intentional states, inferring beliefs and desires from its consistently rational pattern of behavior.

Intentional System

An entity that, when viewed from the intentional stance, exhibits a consistent and rational pattern in its behavior, justifying the attribution of intentional states to it.

Derivative Intentionality

Intentionality possessed by artifacts (like sentences, pictures, or AI systems) that is borrowed or derived from the original intentionality of their intelligent users or designers.

Original Intentionality

Intentionality that is not derivative or borrowed, meaning a system possesses genuine understanding and a mind of its own, as implied by Dennett's view of consistently rational behavior.

Formal System

A system (like a game) in which tokens are manipulated according to definite, self-contained rules, specifying how arrangements of tokens may or must be changed into others, based only on the current arrangement.

Token-Manipulation System

A system defined by a set of token types, allowable starting positions, and formal rules for changing arrangements of tokens, all operating without external reference.

Digital System

A technique for making and identifying things where reidentification can be absolutely perfect (positive) and almost always is perfect (reliable), insulating it from slight physical variations.

Medium Independence

A characteristic of concrete systems like formal systems, where their essential nature does not depend on the physical material ('medium') in which they are implemented, provided the medium supports the necessary form/structure.

Automatic Formal System

A physical device or machine that automatically manipulates configurations of its parts or states (regarded as tokens and positions) in strict accord with the rules of some formal system (e.g., computers).

Implementation (in Computer Science)

Using one automatic formal system to make another, via programming, where a 'general purpose' computer acts as if it were a 'special computer' by treating some of its internal configurations as tokens and positions of the latter.

Universal Machine (Turing)

A theoretical computing machine capable of implementing any well-defined automatic formal system, given sufficient storage and time; every ordinary programmable computer is one.

Algorithm

A procedure that is guaranteed to always work, effecting a desired rearrangement or solving a problem regardless of the input, in a finite amount of time.

Heuristic

A procedure or rule of thumb that is not guaranteed to give the right answer every time but is fairly reliable most of the time, used for problems without known algorithmic solutions (e.g., chess computers).

Digital Simulation

A process where dynamic, non-digital systems (whose states change over time) are represented by dividing time and variables into small intervals and computing their progress step by tiny step, using digital numerals.

GOFAI (Good Old-Fashioned AI)

An early dominant approach to AI that views the mind as essentially a digital computer with internal states and processes regarded as explicit thinking or reasoning, focusing on rational manipulation of internal symbol structures.

Interpreted Formal System

A formal system whose tokens ('symbols') are understood to mean something (e.g., numerals representing numbers), linking formal manipulation with semantic meaning.

Symbol (in Formal Systems)

A token within a formal system that is assigned meaning, often representing external entities, operations, or relationships.

Intentional Interpretation

A species of interpretation that attributes intentional states (like beliefs and desires) to a system or creature based on its behavior.

Semantic Interpretation (Translation)

A species of interpretation that attributes meanings to strings of tokens (symbols) within a system, aiming to construe them as consistently reasonable and sensible.

Holism (of Interpretation)

The principle that it is impossible to interpret a brief string of symbols or a single bit of behavior completely out of context; meaning or rationality is inferred from consistent patterns.

GOFAI Common Sense Problem

The limitation of early GOFAI systems that, despite clever solutions to carefully posed problems, they lacked broad 'common sense' knowledge and were prone to obvious blunders in unexpected variations.

Knowledge Representation (GOFAI)

A GOFAI research focus on organizing and structuring vast amounts of common-sense knowledge efficiently so that a system can access and use it to avoid mistakes and handle ambiguity.

NFAI (New-Fangled AI)

A collective term for alternative approaches to scientific mind design that are not GOFAI, including connectionism and embodied/embedded AI.

Connectionism

A prominent NFAI approach involving networks of simple active units with many weighted connections between them, that interact to process patterns collectively, without a central processor or separate memory.

Connectionist Memory (Short-Term)

Information retained in a connectionist system over time due to units changing state slowly and regularly.

Connectionist Memory (Long-Term)

Information stored in the 'weights' or 'strengths' of connections between units, which are preserved over time and determine the network's basic character.

Active Units (Connectionism)

Simple components in a connectionist network (analogous to neurons) whose 'activation level' (a single quantitative magnitude) changes in response to summed-up signals from other units, processed through connection weights.

Pattern Processors (Connectionist Networks)

The essential function of connectionist networks, making them adept at finding similarities, recognizing repeated patterns, filling in missing parts, and transforming patterns rapidly and in parallel.

Training Connectionist Networks

A method where networks adjust their connection weights through exposure to sample instances, enabling them to acquire desired pattern-processing abilities.

Embodied and Embedded AI

An NFAI approach suggesting that an intelligent system is not just an internal subsystem (like a brain or computer) but a larger, interactive whole, including an active body and environment as essential components for intelligence.

Understanding (Haugeland's definition)

A missing element in current AI, characterized by a system taking responsibility for correctly applying its proto-concepts, for the empirical adequacy of the proto-concepts themselves, and for taking a firm stand on what can and cannot happen in the world, grasping truth/falsity for itself.

Proto-Concept

Related distinctions made by a system in coping with objects, which can manifest as skillful responses or symbol structures, that precede full conceptual understanding.