The Society of Mind
Author calls Society of Mind the scheme in which each mind is made of many smaller processes. These he calls agents. Each mental agent by itself can only do some simple thing that needs no mind or thought at all. Yet when we join these agents in societies – in certain very special ways – this leads to true intelligence.
Agents of the mind
- Function: How do agents work?
- Embodiment: What are they made of?
- Interaction: How do they communicate?
- Origins: Where do the first agents come from?
- Heredity: Are we all born with the same agents?
- Learning: How do we make new agents and change old ones?
- Character: What are the most important kinds of agents?
- Authority: What happens when agents disagree?
- Intention: How could such network want or wish?
- Competence: How can group of agents do what separate agents cannot do?
- Selfness: What gives them unity or personality?
- Meaning: How could they understand anything?
- Sensibility: How could they have feelings or emotions?
- Awareness: How could they be conscious or self-aware?
The mind and the brain
We need better theories about how thinking works.
The society of mind
You know that everything you think or do is thought and done by you. But what’s you?
In science, one can learn the most by studying what seems the least.
The world of blocks
Imagine a child playing with blocks, and imagine that this child’s mind contains a host of smaller minds. Call them mental agents.
Builder is working with Add, Find, Get, Put.
Common sense
Common sense is not a single thing. Instead, it is an immense society of hard-earned practical ideas – of multitudes of life-learned rules and exceptions, dispositions and tendencies, balances and checks.
Agents and agencies
Intelligence as a combination of simpler things.
Whenever we find that an agent has to do anything complicated, we’ll replace it with a subsociety of agents that do simpler things.
It is not enough to explain only what each separate agent does. We must also understand how those parts are interrelated – that is, how groups of agents can accomplish things.
Agent and agency. As agency, it seems to know its job. As agent, it cannot know anything at all.
Knowing how is not the same as knowing why.
Wholes and parts
Components and connections
Find, we must know how each separate part works. Second, we must know how each part interacts with those to which it is connected. And third, we have to understand how all these local interactions combine to accomplish what that system does – as soon from the outside.
Novelists and reductionists
It’s always best when mysteries can be explained in terms of things we know.
Reductionists these people who prefer to build on old ideas, and Novelists the ones who like to champion new hypotheses.
Parts and wholes
We’re often told that certain wholes are more than the sum of their parts. We hear this expressed with reverent words like holistic and gestalt.
Sometimes giving names to things can help by leading us to focus on some mystery. It’s harmful, though, when naming leads the mind to think that names along bring meaning close.
Holes and parts
Words like living and thinking are useful for describing phenomena that result from certain combinations of relationships.
Easy things are hard
We’re least aware of what our minds do best.
Are people machines?
Knowing how to use something is not the same as knowing how it works.
Conflict and compromise
Conflict
I will assume that conflicts between agents tend to migrate upward to higher level.
Noncompromise
To settle arguments, nations develop legal systems, corporations establish policies, and individuals may argue, fight, or compromise-or turn for help to mediators that lie outside themselves.
The Principle of Noncompromise: The longer an internal conflict persist among an agent’s subordinates, the weaker becomes that agent’s status among its own competitors. If such internal problems aren’t settled soon, other agents will take control and the agents formerly involved will be dismissed.
Tiny mental agents simply cannot know enough to be able to negotiate with one another or to find effective ways to adjust to each other’s interference. Only larger agencies could be resourceful enough to do such things.
Hierarchies
Designing any society, be it human or mechanical, involves decisions like these:
- What agents choose which others to do what jobs?
- Who will decide which jobs are done at all?
- Who decide what efforts to expend?
- How will conflicts be settled?
Heterarchies
A hierarchical society is like a tree in which the agent at each branch is exclusively responsible for the agents on the twigs that branch from it.
But hierarchies do not always work.
Memories can be defined in terms of keeping records of the past. Agencies also need other kind of memory as well.
Destructiveness
What happens if different agents compete between themselves and take over.
Pain and pleasure simplified
When you are in pain, it’s hard to keep your interest in other things. Pain simplifies your point of view. The same with pleasure. It also simplifies your point of view.
Pain interferes with making plans by undermining interest in anything that’s not immediate.
In order to appear opposed, two things must serve related goals – or otherwise engage the selfsame agencies.
The self
The self
We all believe that human minds contain those especial entities we call selves. But no one agrees about what they are.
Only in logic and mathematics do definitions ever capture concepts perfectly.
Our ideas about our Selves include: person0s self-image and person’s self-ideals.
One self of many?
Every mind contain some sort of Voyeur-Puppeter inside.
It is about single-self view or multiple-self view.
The soul
The value of a human self lies not in some small, precious core, but in its vast, constructed crust.
What are those old and fierce beliefs in spirits, souls, and essences? They’re all insinuations that we’re helpless to improve ourselves.
The conservative self
To understand how something works, one has to know its purpose.
To understand what we call Self, we first must see what Selves are for. One function of the Self is to keep us from changing too rapidly.
Exploitation
There must be checks and balances. We’d never get through one full day if any agency could seize and hold control over all the rest.
Most of our self-control methods proceed unconsciously, but we sometimes resort to conscious schemes in which we offer rewards to ourselves.
If self-control were easy to obtain, we’d end up accomplishing nothing at all.
Self-control
We often perform actions that change the brain’s chemical environment.
Long-range plans
In order to commit ourselves to our largest, most ambitious plans, we learn to exploit agencies that operate on larger spans of time.
Which are our slowest-changing agencies of all? Silent, hidden agencies that shape what we call character. These are the systems that are concerned not merely with the things we want, but with what we want ourselves to be – that is, the ideals we set for ourselves.
Ideals
We usually reserve the word ideals to refer to how we think we ought to conduct our ethical affairs. But author will use the term in broader sense, to include the standards we maintain, for how we ought to think about ordinary things.
Without enduring self-ideals, our lives would lack coherence. A working society must evolve mechanisms that stabilize ideals.
Individuality
Circular causality
We like to explain things in terms of simple cause and effect. But in real life the causal relations between feelings and thoughts are rarely so simple.
We’re always enmeshed in causal loops.
Unanswerable question
What is the purpose of life? What caused universe, and why? How can you tell which beliefs are true? How can you tell what is good?
You can never find a final cause, since you must always ask one question more: “What caused the cause?”
All human cultures evolve institutions of law, religion, and philosophy, and these institutions both adopt specific answers to circular questions and establish authority-schemes to indoctrinate people with those beliefs.
The remote-control self
What controls the brain? The mind.
What controls the mind? The self.
What controls the Self? Itself.
Personal identity
Why do we accept that paradoxical image of a central Self inside the self? Because it serves us well in many spheres of practical life. Here are some reasons to regard a person as a single thing.
- The Physical World: Two people cannot fit where there is room for only one.
- Personal Privacy: Without the concept of the individual. We could have no sense of responsibility.
- Mental Activity: We often find it hard to think two different thoughts at once.
Fashion and style
Recognizability, uniformity, predictability. It can save a lot of mental work if one makes each arbitrary choice the way one did before.
Edward Fredkin’s Paradox: The more equally attractive two alternatives seem, the harder it can be to choose between them.
Traits
Why should our personalities show such coherencies? How could it be that a system assemble from a million agencies can be described by short and simple strings of words?
- Selectivity.
- Style.
- Predictability.
- Self-Reliance.
Permanent identity
So far as consciousness is concerned, we find it almost impossible to separate the appearances of things from what they’ve come to mean to us.
Insight and introspection
Consciousness
In every normal person’s mind there seem to be some processes that we call consciousness.
We gave the name signals to acts whose consciousness are not inherent in their own character but have merely been assigned to them.
Signals and signs
How do we ever understand anything? Almost always by using one or another kind of analogy – that is by representing each new thing as though it resembles something we already know.
We tend to think of knowledge as good in itself, but knowledge is useful only when we can exploit it to help us reach our goals.
Thought-experiments
How do we discover things about our mind? We use a similar technique. We make up little bits of theories about how we think, then test them with tiny experiments.
Thinking affects our thoughts.
To find out what’s happening, programmers have developed special programs for “debugging” other programs.
We cannot handle interruptions perfectly. This doesn’t mean that consciousness cannot be understood in principle. It only means that to study it, we’ll have to use the less direct method of science, because we cannot simply look and see.
B-brains
Connect the A-brain’s inputs and outputs to the real world – so it can sense what happens there. But don’t connect the B-brain to the outer world at all; instead, connect it so that the A-brain is the B-brain’s world.
The B-brain could do experiments with the A-brain, just as the A-brain can experiment with the body or with the objects and people surrounding it.
Frozen reflection
No supervisor can know everything that all its agents do.
Good scientists never try to learn to much at once. Instead, they select particular aspects of a situation, observe carefully, and make records.
Using the mind to examine itself is like science in another way. Just as physicists cannot see the atoms they talk about, psychologists can’t watch the processes they try to examine.
Momentary mental time
Each agency must have at least a slightly different sense of what had happened in the past – and of what is happening now. Each different agent of the mind lives in a slightly different world of time.
The causal now
Our memories are only indirectly linked to physical time. We have no absolute sense of when a memorable event actually happened. At best, we can only know some temporal relations between it and certain other events.
Thinking without thinking
Many people seem absolutely certain that no computer could ever be sentient, conscious, self-willed, or in any other way aware of itself.
The evidence that we are self-aware – that is, that we have any special aptitude for finding out what’s happening inside ourselves – is very weak indeed.
Most of the understanding we call insights are merely variants of our other ways to figure out what’s happening.
Heads in the clouds
We’ll take the view that nothing can have meaning by itself, but only in relation to whatever other meanings we already know.
Later we’ll see how our conceptions of space and time can be based entirely on networks of relationships, yet can still reflect the structure of reality.
Worlds out of mind
There is no singularly real world of thoughts; each mind evolves its own internal structure.
In mental realms, we make up countless artificial schemes to force things to seem orderly, by specifying legal codes, grammar rules.
Some accidents of mental stress has temporarily suppressed the capacity to question, doubt, or probe.
In-sight
Where do we get ideas we need? Most of our concepts come from the communities in which we’re raised.
Brains don’t manufacture thoughts in the direct ways that muscles exert forces or ovaries makes estrogens; instead, to get a good idea, one must engage huge organizations of submachines that do a vast variety of jobs.
Internal communication
We overestimate how much we actually communicate.
The words and symbols we use to summarize our higher-level goals and plans are not the same as the signals used to control lower-level ones.
Meaning itself is relative to size and scale; it takes sense to talk about a meaning only in a system large enough to have many meanings.
Self-knowledge is dangerous
In ordinary life, our pleasure systems help us learn – and therefore, to behave ourselves – by forcing checks and balances on us.
If we could deliberately seize control of our pleasure systems, we could reproduce the pleasure of success without the need for any actual accomplishments. And that would be the end of everything.
Confusion
It’s mainly when our systems fail that consciousness becomes engaged.
Problems and goals
Intelligence
Many people insist on having some definition of intelligence. Intelligence can mean the ability to solve hard problems.
Intelligence is our name for whichever of those processes we don’t yet understand.
Uncommon sense
To be considered an expert, one needs a large amount of knowledge of only a relatively few varieties. In contrast, an ordinary person’s common sense involves a much larger variety of different types of knowledge – and this requires more complicated management systems.
The puzzle principle
We can program a computer to solve any problem by trial and error, without knowing how to solve it in advance, provided only that we have a way to recognize when the problem is solved. It is only two-part program. Generate and test.
Problem solving
The Progress Principle: Any process of exhaustive search can be greatly reduced if we posses some way to detect when progress has been made.
Goals and Subgoals: The most powerful way we know for discovering how to solve a hard problem is to find a method that splits it into several simple ones, each of which can be solved separately.
Using Knowledge: The most efficient way to solve a problem is to already know how to solve it. Then one can avoid search entirely.
We must acquire knowledge, represent it and then develop a process to exploit it.
Learning and memory
There is an old and popular idea that we learn only what we are rewarded for. Reinforcement by reward.
But how do we create new ideas. Learning the better way we learn.
Reinforcement and reward
In order to solve complicated problems, any machine of limited size must be able to reuse its agents in different ways in different contexts.
Local responsibility
The Local scheme rewards each agent that helps accomplish its supervisor’s goal. The Global scheme rewards only agents that help accomplish top-level goals.
Difference-engines
Whenever we talk about goals, we mix a thousand meanings in one word.
A goal-driven system does not seem to react directly to the stimuli or situations it encounters. Instead, it treats the things it finds as objects to exploit, avoid, or ignore.
A difference-engine must contain a description of a “desired” situation.
Actual input and ideal inputs lead to situation and goal description, differences between the two is then transferred by agents.
Intentions
One ingredient of having a goal is persistence. The other critical ingredient of goal is to have some image or description of a wanted or desired state.
The difference-engine scheme remains the most useful conception of goal, purpose, or intention yet discovered.
Genius
Author suspects that genius needs one thing more; in order to accumulate outstanding qualities, one needs unusually effective ways to learn.
A theory of memory
K-Lines: A theory of memory
How is knowledge represented? How is stored? How is it retrieved? How it is used?
Knowledge line or K-line. We keep each thing we learn close to the agent that learn it in the first place.
Kenneth Haase had a great deal of influence on this theory.
Re-membering
We have old agents and new agents, the one that are aroused by similarity.
Mental states and dispositions
The experience we found the easiest to recollect are often just the kinds we find the hardest to describe. We cannot always judge the complexity of our mental states by how easily we can express them in words.
We cannot always judge the complexity of our mental states by how easily we can express them in words.
Once we think in terms of K-line memories, it becomes easy to imagine, at least in principle, how a person could recall a general impression of a complex previous experience.
Partial mental states
We make our new ideas by merging parts of older ones.
A total state of mind is a list that specifies which agents are active and which are quiet at a certain moment.
A partial state of mind merely specifies that certain agents are active but does not say which other agents are quiet.
Level-bands
We learn by attaching agents to K-lines, but we don’t attach them all with equal firmness. We make strong connections at a certain level of detail, but we make weaker connections at higher and lower levels.
Let’s call weakly activated memories assumptions by default.
Levels
Levels of memories.
Fringes
It’s hard to recognize a thing when you’re presented with too much detail.
Because in order to think, we need intimate connections between things and goals – between structures and their functions.
Societies and memories
Our trick of connecting new K-lines to old ones will not recapture so many of the scene’s precise, perceptual memories.
Knowledge-trees
If each K-line can connect to other K-lines, which, in turn, connect to others, then K-lines can form societies.
This policy of connecting new K-lines to old ones must be used in moderation. Otherwise, no new agents would ever be included in our memories.
Levels and classifications
Isn’t it interesting how often we find ourselves using the idea of level?
If theory of K-lines trees is correct, it would seem natural for us to classify things into levels of hierarchies.
Layers of societies
Original S-agents form S-society. K-society is constructed from S-society which came first.
Summaries
Wanting and liking
What makes us want to compress so much into such inexpressive summaries as like, prefer, enjoy.
To choose between alternatives, the highest levels of the mind demand the simplest summaries.
The relation between wanting and liking is not simple at all, because our preferences are the end products of so many negotiations among our agencies.
Gerrymandering
The only way to solve hard problems is by breaking them into smaller ones and then, when those are too difficult, dividing them in turn. So hard problems always lead to branching threes of subgoals and subproblems.
Learning from failure
Whenever you try to improve an already working procedure, you risk damaging whichever other skills depend on the same machinery.
Perfect logic rarely works in the real world of people, thoughts, and things.
Enjoying discomfort
There is more motivation than immediate reward. When we succeed at anything, a lot goes on inside the mind.
In the early stages of acquiring any really new skill, a person must adopt at least a partly antipleasure attitude.
Papert’s principle
Piaget’s experiments
Jean Piaget realized that watching children might be a way to see how mind-societies grow.
Reasoning about amounts
The younger children posses the ideas they need but don’t know when to apply them. They lack the adequate knowledge about their knowledge.
Priorities
People indeed seem most sensitive to vertical extents. We do not know whether this is built from the start into our brains, but in any case, the bias is usually justified because more height so frequently goes along with other sorts of largenesses.
Papert’s principle
What should one do when different kinds of knowledge don’t agree.
Seymour Papert work from 1960 is talking about mind developing not only with knowledge accumulation but also its usage.
Papert’s Principle: Some of the most crucial steps in mental growth are based not simply on acquiring new skills, but on acquiring new administrative ways to use what one already knows.
The society-of-more
Each higher-level agent embodies a form of higher-order knowledge that helps us organize ourselves by telling us when and how to use the things we know.
About Piaget’s experiments
Piaget used his experiments to show that children are not small adults.
The concept of concept
We’re imprisoned by our poverty of words because even though we have good ways to describe objects and actions, we lack methods for describing dispositions and processes.
Thing-ifying is indeed a splendid mental instrument.
Education and development
It is fairly easy to resolve conflicts by switching among alternatives. It is much harder to develop mechanisms that can use cooperation and compromise – because that requires more complex ways for agencies to interact.
Learning a hierarchy
How could a brain continue functioning while changing and adding new agents and connections? One way would be to keep each old systems unchanged while building a new version in the form of a detour around or across it – but not permitting the new version to assume control until we’re sure that it can also perform the older system’s vital functions.
The shape of space
Seeing red
What possible kind of brain-event could correspond to anything like the meaning of an ordinary word?
The shape of space
We never actually make any direct contact with the outside world. Instead, we work with models of the world that we build inside our brains.
Scientists know a good deal about how these sensors send signals to the brain. But we know much less about how those signals lead to sensations of touch and of sight.
Nearnesses
The reason our skin can feel is because we’re built with myriad nerves that run from every skin spot to the brain.
Innate geography
We’ve seen that touching nearby spots of skin will usually give rise to similar sensations.
Sensing similarities
What we learn depends on how we classify.
Our genes supply our bodies with many kinds of sensors – external even-detecting agents.
The centered self
How do we learn about the real, three-dimensional world?
Critical step would be developing some agents that represent a few places outside the skin. Once those places are established (the first ones might be near the infant’s face), one could proceed to another stage.
Predestined learning
It would be wonderful if we could classify all behavior into two types: built-in and learned. Heredity and environment.
Perhaps the growth of the Society-of-More is another instance of predestined learning.
Half-brains
Our brains have many pairs of agencies, arranged like mirror-images, with huge bundles or nerves running between them.
Dumbbell theories
Dividing things in two is a good way to start, but one should always try to find at least a third alternative. If one cannot, one should suspect that there may not be two ideas at all, but only one, together with some form of opposite.
Learning meaning
A block-arch scenario
Our child, playing with some blocks and toy car, happens to build this structure. Let’s call it a Block-arch.
Learning meaning
Learning is making useful changes in the workings of our minds.
One Block-Arch scene reveals at least four different ways to learn: uniframing, accumulating, reformulating, trans-framing.
Uniframes
The new word uniframe – a description constructed to apply to several different things at once.
How can we judge which facts are useful? We simply have to learn a large society of different ways to learn.
Structure and function
To learn new words or new ideas, one must make connections to other structures in the mind.
Much of what we think in later life is based on what we learn in early life about the world of space.
The functions of structures
Many things that we regard as physical are actually psychological.
Until we learn to make old descriptions fit new circumstances, our old knowledge can be applied only to the circumstances in which it was learned. And that would scarcely ever work, since circumstances never repeat themselves perfectly.
Accumulation
Uniframing doesn’t always work. We often try to make an everyday idea precise – but just can’t find much unity. Then, we can only accumulate collections of examples.
Accumulation need not take longer to manipulate it all the examples can be handled at the same time, by separate agents that don’t interfere with one another.
Accumulation strategies
People personalities:
Uniframers: disregard discrepancies in favor of imagined regularities. They tend to be perfectionists but also tend to think in terms of stereotypes.
Accumulators: are less extreme. They keep collective evidence and hence are much less prone to make mistakes.
Problems of disunity
When should you accumulate and when should you make uniframes? The choice depends on your purposes.
Whenever we build a bridge between structure and function, one end of that bridge may represent a goal or use, while the other end describes what we might use to gain those ends. We usually find many different ways to achieve any goal.
The exception principle
The Exception Principle: It rarely pays to tamper with a rule that nearly always works. It’s better just to complement it with and accumulation of specific exceptions.
We almost never find rules that have no exceptions.
To insist on perfect laws in real life is to risk not finding any laws at all.
How towers work
We often use words like insight or intuition to talk about understanding that seem especially immediate. But it is bad psychology to assume that what seems obvious is therefore simple or self-evident.
How causes work
Why can we explain so many things in terms of causes and effects? It is because there is cause for everything.
There can’t be any causes in a world in which everything that happens depends more or less equally upon everything else that happens.
To know the cause of a phenomenon is to know, at least in principle, how to change or control some aspect of some entities without affecting all the rest.
Meaning and definition
Meaning – that what exists in the mind. That which is intended to be by acts or language, the sense, signification, or import of words.
Bridge-definitions
Purposeful definitions are usually too loose. They include many things we do not intend.
Structural definitions are usually too high. They reject many things we want to include.
Seeing and believing
Reformulation
Reformulation is clearly very powerful – but how does one do it? How do people find new styles of description that make their problems seem easier?
Boundaries
What is creativity? How do people get new ideas? Most thinkers would agree that some of the secret lies in finding new ways to look at things.
We are always changing boundaries.
Seeing and believing
We normally assume that children see the same as we do and only lack our tricky muscle skills.
Children’s drawing-frames
It does not make much sense to speak of what person really sees, because we have so many different agencies.
In order to make mature, realistic drawings, the child could exploit the same kind of ability it must acquire in order to count things properly.
Learning a script
An expert is one who does not have to think. He knows. – Frank Lloyd Wright[1]
Perhaps when we practice to improve our skills, we’re mainly building simpler scripts, that don’t engage so many agencies. This lets us do old things with much less thought and gives us more time to think of other things.
The frontier effect
Young children don’t yet possess much ability to draw lines in good proportion.
Piaget called this the frontier effect. The tendency to place new features at locations that have easily described relationships to other, already represented features.
It can require more skills to produce what we regard as a simple copy or imitation than to produce what we consider to be an abstract representation.
Duplications
We could also formulate this as a choice between structural and functional descriptions.
The functional type of description is easier to adapt to the purposes of higher-level agencies. This does not mean that they are better.
Reformulation
Using reformulations
What we can do when we can’t solve a problem? We can try to find a new way to look at it, to describe it in different terms. Reformulation is the most powerful way to attempt to escape from what seems to be a hopeless situation.
How do we reformulate? Each new technique presumably begins by exploiting methods already learned in other, older agencies. So new ideas often have roots in older ones, adapted for new purposes.
The body-support concept
We were able to uniframe many kinds of arches by dividing each into a Body and a Support.
The body-support representation helps us to classify our knowledge. The body represents these parts of a structure that serve as the direct instrument for reaching the goals and the support represents all the other features that merely serve the instrument.
Means and ends
How do we connect the things we have with the goals we want to achieve.
Our systematic cross-realm translations are the roots of fruitful metaphors; they enable us to understand things we’ve never seen before.
Seeing squares
We often self-impose assumptions that make our problems more difficult and we can escape from this only reformulating those problems in ways that give us more room.
The way we perceive the world, from one moment to another, depends only in part on what comes from our eyes. The rest of what we think we see comes from inside our brain.
Human vision must somehow combine the information that comes from the outer world with the structures in our memories.
Brainstorming
Remember the principle of exceptions; it may be rash to change yourself too much just to accommodate a single strange experience.
The investment principle
The Investment Principle: Our oldest ideas have unfair advantages over those that come late. The earlier we learn s skill, the more methods we can acquire for using it. Each new idea must then compete against the larger mass of skills the old ideas accumulated.
Parts and holes
An obstacle is an object that interferes with the goal of moving in a certain direction.
The power of negative thinking
How do boxes keep things in? The top-level agent Move has four subagents: Move-Left, Move-Right, Move-Up and Move-Down. If every direction is blocked, the arm can’t move at all.
The interaction-square
Many of our body joints can move in two independent direction at once – not the knee, but certainly the writs, shoulder, shoulder, hip, ankle, thumb, and eye.
Rarely, it seems, do people deal with more than two causes at a time; instead we either find ways to reformulate such situations or we accumulate disorderly societies of partially filled interaction-squares that cover only the most commonly encountered combinations.
Consciousness and memory
Momentary mental state
Author will argue that consciousness does not concern the present, but he past; it has to do with how we think about the records of our recent thoughts.
There is nothing peculiar about the idea of sensing events inside the brain.
Only a small minority of our agents are connected directly to sensors in the outer world, like those that send signals from the eye or skin; most of the agents in the brain detect events inside the brain.
Self-examination
Why is so hard to talk about our present state of mind? One reason is time-delays between different parts of mind and the concept of present state of mind that is not psychologically sound. The second reason is that ever reflection changes the state.
Memory
In order for a mind to think, it has to juggle fragments of its mental states.
We have two different mechanisms: long-term and short-term memories.
Our various agencies selectively decide to transfer only certain states into their long-term memories.
Memories of memories
It’s hard to distinguish memories from memories of memories.
Memories are processes that make some of our agents act in much the same way they did at various times in the past.
The immanence illusion
Memories can’t really bring things back; they only reproduce some fragments of our former states of mind.
The Immanence Illusion: Whenever you can answer a question without a noticeable delay, it seems as though that answer were already active in your mind.
Our sense of momentary mental time is flawed; our vision-agencies begin arousing memories before their own work is fully done.
Many kinds of memory
Everyone has many kinds of memories.
A brain has no single, common memory system. Instead, each part of the brain has several types of memory-agencies that work in somewhat different ways, to suit particular purposes.
Memory rearrangements
We have to be able to connect our short-term memories only to appropriate aspects of our current problems. Learning such abilities is not simple, and perhaps it is a skill some people never really master.
Anatomy of memory
No one yet knows how memories control themselves inside our brain.
There is good evidence that, in human brains, the process that transfer information into long-term memory are very slow, requiring time intervals that range from minutes to hours. Accordingly, most temporary memories are permanently lost.
Interruption and recovery
The problem of keeping track of what is happening is hard enough when one agency calls on another one for help. Until the other’s job is done, the first agency has to keep some temporary record of what it was doing.
Interruption is a problem because we have to keep our place in several processes at once.
Losing track
Whenever we solve complicated problems, we get into situations in which our agencies must keep account of many processes at once. It can be called stack. We simply aren’t very good at dealing with the kinds of situations that need such memory-stacks.
The recursion principle
The Recursion Principle: When a problem splits into smaller parts, then unless one can apply the mind’s full power to each subjob, one’s intellect will get dispensed and leave less cleverness for each new task.
Emotion
Emotion
Why do so many people think emotion is harder to explain than intellect?
The use of fantasies, emotional or not, is indispensable for every complicated problem-solving process. We always have to deal with nonexistent scenes.
No matter how neutral and rational a goal may seem, it will eventually conflict with other goals if it persist for long enough.
The question is not whether intelligent machines can have any emotions, but whether machines can be intelligent without any emotions.
Mental growth
How do our minds form? Is every person born containing a hidden, built-in intellect just waiting to reveal itself? Or must minds grow in little steps from emptiness?
Human minds don’t merely learn new ways to reach old goals, we can also learn new kinds of goals.
Mental proto-specialists
Proto-specialists. Each has a separate mini-mind to do its job and is equipped with special sensors and effectors designed to suit its specific needs.
Cross-exclusion
If several urgent needs occur at once, there must be a way to select one of them. One way is a central market place that decides out of competing needs.
An arrangement called cross-exclusion, which appears in many portions of the brain. In such a system, each member of a group of agents is wired to send inhibitory signal to all the other agents of that group. This makes them competitors.
Avalanche effects
What prevents our brains from avalanches of activity. The cross-exclusion scheme is one way. Other are: conservation, negative feedback and censors and suppressors.
Motivation
Whenever any specialist tried to rearrange some memories to its own advantage, it might damage structures upon which the other have come to depend.
Exploitation
Achieving a goal by exploiting the abilities of other agencies might seem a shabby substitute for knowing how to do the work oneself. Yet this is the very source of the power of societies.
Stimulus vs. stimulus
Apparently one agency can activate another merely by imagining a stimulus.
A stimulus is a reproduction of only the higher-level effect of a stimulus.
Infant emotions
How should we interpret an infant’s apparent single-mindedness? One explanation of those striking shifts in attitude is that one agency attains control of forcibly suppresses the rest. Another view is that many processes continue at once – but only one at a time can be expressed.
Adult emotions
Our earliest emotions are built-in processes in which inborn proto-specialists control what happens in our brains.
Development
Sequences of teaching-selves
It is usually easier to recognize a solution to a problem than to discover a solution; this is what we called the puzzle principle.
Attachment-learning
In the case of ordinary forms of failure or success signals, the learner modifies the methods used to reach the goals. In the cased of fear-provoking disturbances, the learner may modify the description of the situation itself. In the cased of attachment-related failures or reward signals, the learner modifies which goal are considered worthy of pursuit.
Attachment simplifies
In many species attachment occurs naturally. It may be called imprinting.
Attachments teach us ends, not means – and thus, impose on us our parents’ dreams.
Functional autonomy
Functional Autonomy: In the course of pursuing any sufficiently complicated problem, the subgoals that engage our attentions can become both increasingly more ambitious and increasingly detached from the original problem.
Knowledge is power.
Virtually any problem will be easier to solve the more one learns about the context world in which that problems occurs.
Developmental stages
Author will argue that nothing so complex as a human mind can grow, except in separate steps.
One conservative strategy is never to let a new stage take control of actual behavior until there is evidence that it can outperform its predecessor.
Prerequisites for growth
From time to time, each parent has the illusion that a child has suddenly changed, when this is only the result of not observing several smaller, real changes in the past.
Genetic timetables
Once it becomes too hard to change an old agency, it is time to build another one, further progress may require revolution rather than evolution.
Attachment-images
Consider that our models of ourselves are so complex that even adults can’t explain them.
Many people dislike the thought of being dominated from within by the image of a parent’s wish.
Different spans of memories
Some memories are less changeable than others, and author suspect that attachment-bonds involve memory-records of a type that can be rapidly formed but then become peculiarly slow to change.
Intellectual trauma
By itself, the failure to achieve a goal can cause anxiety. In early life such an incident could have produced a crisis of self-confidence and helplessness.
We generally accept incompetence of intellect as a normal, if unfortunate, deficiency in talents, aptitudes, and gifts.
Intellectual ideals
Human thought is not based on any single and uniform kind of logic, but upon myriad processes, scripts, stereotypes, critics and censors, analogies and metaphors.
Reasoning
Must machines be logical
Without an intimate connection between our knowledge and our intentions, logic leads to madness, not intelligence.
Chains of reasoning
We often think in terms of causes, similarities, and dependencies. What do all these forms of thinking share? They all use different ways to make chains.
Chaining
Chaining seems to permeate not only how we reason, but how we think of structures in space and time.
Logical chains
Logic is the word we use for certain ways to chain ideas. Logic demands just one support for every link, a single, flawless deduction. Common sense asks, at every step, if all of what we’ve found so far is in accord with everyday experience.
Strong arguments
In logic, arguments are simply either right or wrong. We can use different methods to make our chain of reasoning harder to break. One method is to use several different arguments to prove the same point – putting them in parallel.
Magnitude from multitude
How to make argument stronger. Strength from magnitude and strength from multitude.
What is a number
What something means depends upon each different person’s state of mind. The closest we can come to agreeing on meaning is in mathematics.
Mathematics made hard
Scientists and philosophers are always searching for simplicity.
In real life, our minds must always tolerate beliefs that later turn out to be wrong.
Robustness and recovery
How could anything be robust: duplication, self-repair, distributed processes and accumulation.
Words and ideas
The roots of intention
Language build things in our minds. Yet words themselves can’t be the substance of our thoughts.
The language-agency
Words are the instrument by which we form all our abstractions.
The language-system is divided into three regions: listener, word-agents and speaker. The language-agency seems to have an unusual capacity to control its own memories.
Words and ideas
Two kinds of agents that contribute to the power of words, the first king, called polynemes, are involved with our long-term memories. Another type we call an isonome. Each isonome controls a short-term memory in each of many agencies.
Objects and prop
We like the kinds of properties that do not change capriciously. The most useful sets of properties are those whose members do not interact too much.
Polynemes
All ideas about meaning will seem inadequate by themselves, since nothing can mean anything except within some larger context of ideas.
Recognizers
The simplest way to recognize something is to verify that it has certain properties.
Weighing evidence
In 1959, Frank Rosenblatt invented an ingenious evidence-weighing machine called a Perceptron.
All feature-weighing machines have serious limitations because, although they can measure the presence or absence of various features, they cannot take into account enough of the relations among the features.
Generalizing
We’re always learning from experience.
There is an intimate relationship between how we represent what we already know and the generalization that will seem most plausible.
Recognizing thoughts
How do we recognize our own ideas? It is not that simple.
Closing the ring
The method for arousing complete recollections from incomplete clues – we could call it reminding – is powerful but imperfect.
Context and ambiguity
Ambiguity
Thoughts themselves are ambiguous.
We can tolerate the ambiguity of words because we are already so competent at coping with the ambiguity of thoughts.
Negotiating ambiguity
Many common words are ambiguous enough that even simple sentences can be understood in several ways.
Visual ambiguity
The point is that what we see does not depend only on what reaches our eyes from the outside world. The manner in which we interpret those stimuli depends to a large extent on what is already taking place inside our agencies.
Locking-in and weeding-out
In a few cycles, the entire system will firmly lock-in on our meaning-sense for each word and firmly suppress the rest.
There is no guarantee that this method will always find an interpretation that yields a meaning consistent with all the words of the sentence.
Micronemes
Author calls micronemes those inner mental context clues that shade our minds’ activities in ways we can rarely express. They participate in looking-in, weeding-out processes.
The nemeic spiral
Our polynemes and micronemes grow into great branching networks. In any case, our higher-level agencies are generally unaware of what our lover-level agents do.
Connections
Any comprehensive theory of the mind must include some ideas about the nature of the connections among agents.
In order to reproduce the major features of a remembered partial state of mind, it should suffice to activate only a representative sample of its agents. Most K-lines’ connections are indirect to begin with, since they connect only to other, nearby K-line trees.
Connection lines
The diagram of connection-scheme that permits many agents to communicate with one another, yet uses surprisingly few connection wires is in the book. The scheme was invented by Calvin E. Mooers in 1946.
Distributed memory
A new type of network machine is so-called Boltzmann machine. It resembles Perceptron – automatic learning of new connection weights and has some ability to resolve ambiguities by using a variety of ring-closing process.
Trans-frames
The pronouns of the mind
Whenever we talk or think, we use pronounlike devices to exploit whatever mental activities have already been aroused, to interlink the thoughts already active in the mind. To do this, we need to have a machinery we can use as temporary handles for taking hold of, and moving around, those active fragments of mental states.
Pronomes
Why are sentences so easy to understand. Typically, an English sentence is built around a verb that represents some sort of act, event or change.
It is about roles and concerns.
Trans-frames
In the early 1970s, Roger Schank developed ways to represent many situations in terms of a relatively few kinds of relations which he called conceptual dependencies. P-trans represents a physical motions from one place to another. M-Trans represents the sort of mental transportation involved.
Communication among agents
How can agents operate without explicit messages?
Automatism
Hod do higher-level agencies tell lower-level agents what to do?
Many of your agencies have become immersed in the context produced by the agents directly involved with whatever subject was mentioned recently.
Focus on mental attention.
Trans-frame pronomes
In order to use chainlike thinking skills, we need to be able to represent what we know in ways that provide connection points.
Trans-like representation-schemes have been very useful in Artificial Intelligence research projects.
Generalizing with pronomes
From every moment to the next, a person’s state of mind is involved with various objects, topics, goals and scripts.
For first few times you try to do something new, you must experiment to find which agents to activate, and at what times, and for how long. Later, you can prepare a script that will do a job more quickly. Two scripts: a pronome-asignment and action script.
Attention
When several object move at once, it’s hard to keep track of all of them. These skills develop over time.
Expression
Pronomes and polynemes
Polynemes are permanent K-lines. They are long-term memories.
Pronomes are temporary K-lines. They are short-term memories.
Isonomes
An isonome has a similar, built-in effect on each of its recipients. It thus applies the same idea to many different things at once. A polynome has different, learned effect on each of its recipients. It thus connects the same thing to many different ideas.
De-specializing
What magic tricks allow us to de-specialize whatever skills we learn? One way to do this is simply replacing certain polynemes with less specific isonomes.
Learning to think in terms of isonomes must be a crucial step in many types of mental growth.
What we call generalizing is not any single process or concept, but a functional term for the huge societies of different methods we use to extend the powers of our skills.
Learning and teaching
The power of what we learn depends on how we represent it our mind.
Some children learn to represent knowledge in versatile ways; others end up with accumulations of inflexible, single-purpose procedures or with almost useless generalities.
Each child learns, from time to time, various better ways to learn – but no one understand how this is done.
Inference
Linking structures together into chains is one of our most useful kinds of reasoning.
Over the years, children improve their abilities to decide when two different structures are similar enough to justify making chain-links.
By learning to manipulate our isonomes, we become able to combine mental representations into structures that resemble bridges, chains and towers.
Expression
Why do we thing-ify our thoughts?
Without the ability to reflect, we would have no general intelligence, however large our repertoire of special-purpose skills might grow.
Causes and clauses
Things, differences, causes, clauses.
Our brain appear to make us seek to represent dependencies.
Interruptions
What enables us to tolerate an interruption and then return to our previous thoughts? This must engage the agents that control our short-term memories.
Pronouns and references
We sometimes think of words like who or it as pronouns – that is, signals that represent or substitute for other nouns or phrases. Communication will fail unless the listener can correctly guess which pronome the speaker wishes to assign to that activity.
Verbal expression
Re-duplication theory of speech describes only the first stages of how we use language.
Creative expression
There is a wonderful capacity that comes along with the ability to express ideas. Whatever we may want to say, we probably won’t say exactly that.
Culture – the conceptual treasures our communities accumulate through history.
Comparisons
A world of differences
There are several kinds of thinking you might do: predicting; expecting; explaining; wanting; escaping, attacking and defending and abstracting.
The ability to consider differences between differences is important because it lies at the heart of our abilities to solve new problems.
Differences and duplicates
The Duplication Problem: The states of two different agencies cannot be compared unless those agencies themselves are virtually identical.
Time blinking
Our sensors react mainly to how things change in time.
Although the method of time blinking is powerful and efficient, it has some limitations, for example, it cannot directly recognize relations among more than two things at a time.
The meanings of more
When we hear the word more, we become disposed to make comparisons.
It is one thing to find a difference, but another to know whether to call it more or less.
Foreign accents
There is evidence that our brains use different machinery for recognizing language sounds than for recognizing other sorts of sounds.
Frames
The speed of thought
Our idea is that each perceptual experience activates some structure that we’ll call frames – structures we’ve acquired in the course of previous experience.
Frames of mind
A frame is a sort of skeleton somewhat like an application form with many blanks or slots to be filled. We’ll call these blanks its terminals, we use them as connection points to which we can attach other kinds of information.
Virtually any kind of agent can be attached to a frame-terminal.
Default assumptions fill our frames to represent what’s typical.
Frames are drawn from past experience and rarely fit new situations perfectly. We have to learn to adapt them to specific situations.
How trans-frames work
One of the great surprises of modern computer science was the discovery that so much can be done with so few kinds of ingredients.
Default assumptions
Default assumptions make weak images, and they usually turn out to be wrong.
Each reader reads only what is already inside himself. A book is only a sort of optical instrument which the writer offers to let the reader discover in himself what he would not have found without the aid of the book.
Nonverbal reasoning
We do this kind of reasoning by manipulating our memories to replace particular things by typical things.
To reason well, our memory-control agencies must learn to move our memories around as thought those memories were building blocks.
Direction-nemes
No one yet know how shapes and places are represented in the brain.
How can we use so many different kinds of information at once?
We represent directions and places by attaching them to a special set of pronome-like agents that we shall call direction-nemes.
Picture-frames
Whenever we see a thing so utterly new that it resembles nothing we’ve ever seen before, this means that none of our prelearned frames will fit it very well.
How picture-frames work
To make a picture-frame, we’ll simply replace the pronomes of our Trans-frame scheme by a set of nine direction-nemes.
Each time you look in a different direction, your vision-system will describe what you see – and the corresponding K-line will record what you see when you look in that direction.
Recognizers and memorizers
How do frames become activated? Every frame is activate by some set of recognizers.
Frame and agencies might be organized in the form of agents sandwiched between recognizers and memorizers.
Frame-arrays
One frame at a time
What enables us to see those pictures as composed of the features we call by names like edges, lines, corners, and areas. What enables us to see those features as grouped together to form a larger objects? Locking-in machinery permits each object to be attached only to one frame at a time.
Frame-arrays
When we move, our vision-systems switch among a family of different frames that all use the same terminals.
The stationary world
When you activate your move east direction-nemes in order to make your body move in that direction, the same signal will also make the frame-array replace the middle frame with the one to its left. This will compensate for your change of viewpoint.
The sense of continuity
Our sense of constant contact with the world is not a genuine experience, instead, it is a form of immanence illustion.
Our sense of smooth progression from one mental state to another emerges not from the nature of that progression itself, but from the descriptions we use to represent it.
Expectations
When we look about a familiar place, we know roughly what to expect. But what does expect means.
The frame idea
The idea was conceived by author in the early 1970s.
Language-frames
Understanding words
We call understandings common sense. They’re made so swiftly that they’re often ready in our minds before a sentence is complete. Such knowledge comes from previous experiences. But it is not simple.
Understanding stories
What makes a story comprehensible? What give is coherence? The secret lies in how each phrase and sentence frames into activity or helps already active ones to fill their terminals.
Sentence-frames
People do not usually state all the parts of a given thought that they are trying to communicate because the speaker tries to be brief and leaves out assumed or unessential information.
A party-frame
Dictionary definitions never say enough.
We take our social customs for granted, as though they were natural phenomena.
Story-frames
Most stories start with just enough to set the scene. Then they introduce some characters, with hints about their principal concerns. Next, the storyteller gives some clues about some main event or problem to be solved.
Sentence and nonsense
Part of what a sentence means depends upon its separate words, and part depends on how those words are arranged.
Grammar is the servant of language, not the master.
Frames for nouns
At various points in their development, most children seem suddenly to comprehend new kind of sentences.
Our language-agents scarcely know what to do with that first string of words (not related words) because it doesn’t fit the patterns we normally use for describing things. These suggests that we use framelike structure for describing nouns as well as verbs – that is, for describing things as well as actions.
Frames for verbs
In the earliest stages of learning to speak, we simply fill the terminals of word-string frames with nemes for words. Then, later, we learn to fill those terminals with other filled-in language-frames.
The policies for assigning phrases to pronomes vary from one language to another.
Language and vision
Did our capacity to deal with phraselike structures evolve first in language or in vision?
Learning language
Each person invents some new ideas, but most of these will die when their owners do, except for those that make their way into the culture lexicon.
If so, perhaps we ought not to wonder so much about how children learn to speak so readily. Instead, we ought to wonder why it takes so long, when they already do so many things inside their heads.
Grammar
There seems scarcely any limit to the complexity of our social inventions for expressing mental processes.
Coherent discourse
Words are merely the external signs of very complex process.
We’re normally quite unaware of how our grammar-tactics constrain us in our choices of words.
Censors and jokes
Demons
How much of the fascination in telling a story, or listening to one, comes from manipulations of our demons’ expectations?
Suppressors
We accumulate memories to tell ourselves what we shouldn’t think.
Suppressor-agents wait until you get a certain bad idea. Then they prevent your taking the corresponding action.
Censor-agents need not wait until a certain bad idea occurs, instead, they intercept the states of mind that usually precede that thought.
Censors
Each censor may, in time, require a substantial memory bank.
Sometimes our censors and suppressors must themselves be suppressed.
Exceptions to logic
We each must learn specific ways to keep from various mistakes.
Exceptions are a fact of life because few facts are always true.
Jokes
Freud said that jokes are stories designed to fool the censors. A joke’s power comes from a description that fits two different frames at once.
Humor and censorship
Humor is involved with how our censors learn.
Laughter
The function of laughing is to disrupt another person’s reasoning.
Humor must have grown along with our abilities to criticize ourselves, starting with simple internal suppressors that evolved into more sophisticated censors.
Good humor
What is good humor?
The mind and the world
The myth of mental energy
Machines and brains require ordinary energy to do their jobs – and need no other, mental forms of energy.
Magnitude and marketplace
We turn to using quantities when we can’t compare the qualities of things.
Quantity and quality
Whenever we turn to measurements, we forfeit some uses of intellect.
Mind over matter
What hurts and even what is felt at all may be more dependent on culture than biology.
The mind and the world
We spend our lives in several realms. This first is ordinary physical world of objects. We also live in a social realm of persons, families and companies. We also live in psychological realm, inhabited by entities we call by names like meanings, ideas, and memories.
Mindsa are simply what brains do.
Minds and machines
The principal activities of brains are making changes in themselves.
Individual identities
Now suppose that we could replace each of your brain cells with a specially designed computer chip that performs the same functions.
Modifying or replacing the physical parts of a brain will not affect the mind it embodies, unless this alters the successions of states in that brain.
Overlapping minds
It makes sense to think that there exists, inside your brain, a society of different minds.
If each of us contains several such mini-minds, could any special exercise help put the all in closer touch?
The realms of thought
The realms of thought
Why is so hard to understand how thoughts relate to things?
Several thoughts at once
We can think in several mental realms at once.
Possession plays essential roles in all our plans, because we can’t use any materials, tools, or ideas until we gain control of them.
Paranomes
Certain pronomes can operate in several different realms at once. Let’s call them paranomes to emphasize their parallel activities.
Whenever any process gains momentary control over a paranome, many other processes can be affected.
Cross-realm correspondences
Language scientists find it hard to classify the roles words play in sentence-frames.
The problem of unity
What makes our minds form many separate mental realms, instead of attempting to see all aspects of the world in a unified way?
To adults, the laws that govern the physical world seem simpler and more orderly than those that apply to human events.
From the point of view of a physically helpless infant, the social realm is by far the simpler one.
Autistic children
Isn’t it curious that infants find social goals easier to accomplish than physical goals, while adults find the social goals more difficult?
Likenesses and analogies
We always try to use old memories to recollect how we solved problems in the past. But nothing is ever twice the same, so recollection rarely match.
Metaphors
Listen closely to anything anyone says, and soon you’ll hear analogies.
We sometimes call these metaphors, our way to transport between the various mental realms.
A metaphor is that which allows us to replace one kind of thought with another.
Good metaphors are useful because they transport uniframes, intact, from one world into another.
Mental models
Knowing
What does knowing really mean?
When we talk about knowledge, don’t we have to say who all those speakers and observers are? Because we make assumptions by default.
Knowing and believing
We often speak as though we classify our thoughts into different types called facts, opinions, and beliefs.
Some philosophers have argued that knowing must mean true and justified beliefs.
Mental models
We all make models of ourselves and use them to predict which sorts of things we’ll later be disposed to do.
World models
When you get right down to it, you can never really describe any worldly thing, either – that is, in any absolute sense. Whatever your purport to say about a thing, you’re only expressing your own beliefs.
Knowing ourselves
Do people go on to make models of their models of their models of themselves? If we kept on doing things like that, we’d get trapped in an infinite regress.
When we find things hard to decide: we can simply say, I just decide.
Freedom of will
Everything that happens in our universe is either completely determined by what’s already happened in the past or else depends on random chance.
Every action we perform stems from a host of processes inside our minds.
My decision is determined by internal forces I do not understand.
The myth of the third alternative
We imagine that somewhere in each person’s mind, there lies a Spirit, Will, or Soul.
No matter that the physical world provides no room for freedom of will: that concept is essential to our models of the mental realms.
Intelligence and resourcefulness
The power of intelligence stems from our vast diversity, not from any single, perfect principle.
The Society of Mind
A gene is a unit of heredity.
Not all genes turn on at once; some start early and some start late.
How do genes encode ideas? The answer lies in the concept of predestined learning.
What possible sort of brain-machine could support a billion-agent society of mind?
The brain of Homo sapiens is mainly composed of cabling.
[1] In the book on page 137
