Notes on 《How the Mind Works》Ch.2 —— Thinking Machine

This chapter mainly deals with 2 questions:

What makes intelligence possible?

Two meanings of "intelligence": aptitude and rational, humanlike thought.

The characteritics of intelligence:

• To make decisions "rationally," by some set of rules (applying a set of operations that reduce the difference.)
• Wanting and pursuing something in the face of obstacles (specifying a goal)
• To use the rational rules to attain the goal in different ways, depending on the obstacles to be overcome. (assessing the current situation to see how it differs from the goal)

Intelligence does not come from a special kind of spirit or matter or energy but from a different commodity, information. Information is a correlation between two things that is produced by a lawful process (as opposed to coming about by sheer chance).

2 properties glued together in the entity we call a symbol:

1. carries information
2. causes things to happen.

The computational theory of mind is the hypothesis that intelligence is computation, which demystified mentalistic terms.

• Beliefs are inscriptions in memory
• desires are goal inscriptions
• thinking is computation
• perceptions are inscriptions triggered by sensors
• trying is executing operations triggered by a goal.

The computational theory of mind also rehabilitates once and for all the infamous homunculus.

1. Homunculi don't duplicate entire the talents they are rung in to explain, instead, a team or committee of relatively ignorant, narrow-minded, blind homunculi to produce the intelligent behavior of the whole.
2. A representation is a set of symbols corresponding to aspects of the world, and each homunculus is required only to react in a few circumscribed ways to some of the symbols.

How a symbol in a mind can mean something:

• a symbol is connected to its referent in the world by our sense organs.
• the unique pattern of symbol manipulations triggered by the first symbol mirrors the unique pattern of relationships between the referent of the first symbol and the referents of the triggered symbols.
• Together the causal and inferential roles of a symbol determine what it represents.

Human brain uses at least four major formats of representation:

1. visual image: like a template in a two-dimensional, picturelike mosaic.
2. phonological representation: a stretch of syllables that we play in our minds like a tape loop. This stringlike representation is an important component of our short-term memory. Phonological short-term memory lasts between one and five seconds and can hold from four to seven "chunks."
3. grammatical representation: nouns and verbs, phrases and clauses, stems and roots, phonemes and syllables, all arranged into hierarchical trees.
4. mentalese: the language of thought in which our conceptual knowledge is couched, the medium in which book's content or gist is captured

Minds is a special case of modular, hierarchical design in all complex systems.Complex systems are hierarchies of modules because only elements that hang together in modules can remain stable long enough to be assembled into larger and larger modules.

Neural networks is more like everything-connected-to-everything network, which sometimes called auto-associator, has five nifty features:

1. a reconstructive, content-addressable memory.
2. use a preponderance of mutually consistent pieces of information to override one unusual piece. "graceful degradation" helps deal with noisy input or hardware failure,
   
3. do a simple version of the kind of computation called constraint satisfaction.
4. ability to generalize automatically.
5. learn from examples, where learning consists of changes in the connection weights.

Two laws governing the thought of the association of ideas (associationism: 观念联想论):

1. contiguity: ideas that are frequently experienced together get associated in the mind. Thereafter, when one is activated, the other is activated too.
   
2. resemblance: when two ideas are similar, whatever has been associated with the first idea is automatically associated with the second.

Five feats giving human thought its distinctive precision and power, but they are the problems for associative network:

1. individuality: neural network has problem to distinguish individuals with identical properties from classes, but human mind can.
2. compositionality: the ability of a representation to be built out of parts and to have a meaning that comes from the meanings of the parts and from the way they are
   combined. Compositionality is the quintessential property of all human languages. Human thoughts are assembled out of concepts; they are not stored whole.
3. quantification (or variable-binding): It arises from a combination of the first problem, individuals, with the second, compositionality. Our compositional thoughts are, after all, often about individuals, and it makes a difference how those individuals are linked to the various parts of the thought. Hooking up concepts to their roles is not enough. Logicians capture these distinctions with variables and quantifiers.
4. recursion: the ability to embed a proposition inside another proposition bestows the ability to think an infinite number of thoughts. Unless neural networks are specially assembled into a recursive processor, they cannot handle our recursive thoughts.
5. People think in two modes, that is, fuzzy and crisp versions of the same category can live side by side in a single head.
   

• People can form fuzzy stereotypes by uninsightfully soaking up correlations among properties, taking advantage of the fact that things in the world tend to fall into clusters
• People can also create systems of rules—intuitive theories—that define categories in terms of the rules that apply to them, and that treat all the members of the category equally.

When the "ideas" were replaced by stimuli and responses, associationism became behaviorism.

What makes consciousness possible?

Different meanings/interpretations of consciousness:

• intelligence:
  
• self-knowledge: an intelligent being can have information about is the being itself.
• access to information: this meanings embraces Freud's distinction between the conscious and the unconscious mind. That is: the mass of information processing in the nervous system falls into two pools:
  
1. One pool, which includes the products of vision and the contents of shortterm memory, can be accessed by the systems underlying verbal reports, rational thought, and deliberate decision making.
   
2. The other pool, which includes autonomic (gut-level) responses, the internal calculations behind vision, language, and movement, and repressed desires or memories (if there are any), cannot be accessed by those systems.
• sentience: subjective experience, phenomenal awareness, raw feels, first-person present tense. The main features of this sense of consciousness: sensory awareness, focal attention, emotional coloring, and the will.

Any intelligent agent incarnated in matter, working in real time, and subject to the laws of thermodynamics must be restricted in its access to information because information has costs:

• space: the hardware to hold the information.
• time:Solving a problem in a hundred years is, practically speaking, the same as not solving it at all.
• resources:Information processing requires energy. working brain tissue calls more blood its way and consumes more glucose.

Access-consciousness has four obvious features:

1. we are aware, to varying degrees, of a rich field of sensation: colors, shapes, sounds, smells, the pressures and aches etc.
• Our immediate awareness seems to tap the intermediate levels, does not exclusively tap the lowest levels of sensations and the highest level of representation, either.
2. portions of this information can fall under the spotlight of attention, get rotated into and out of short-term memory, and feed our deliberative cogitation.
3. sensations and thoughts come with an emotional flavoring: pleasant or unpleasant, interesting or repellent, exciting or soothing.
4. an executive, the "I," appears to make choices and pull the levers of behavior.