Society of Mind
From CasGroup
This page is about the question how the various phenomena of mind emerge from the interactions of many mindless agents. The term Society of Mind has been coined by Marvin Minsky. Bateson's idea of an Ecology of Mind and Minsky's idea of a "Society of Mind" have much in common. Both are similar because they try to describe a whole system of interacting entities - species for ecology and agents for society.
Although consciousness is complicated and confusing, it is like pain/displeasure and joy/pleasure an emergent phenomenon, which arises from the local interactions of billions of agents. It is hard to describe "the whirl of thought and sensation that blossoms when you see a loved one after a long absence, hear an exquisite violin solo, or relish an incredible meal.", see [1], but the general phenomena are maybe very simple.
The Society of Mind
The Chinese room experiment from John Searle says that syntax is not the same as semantics: a symbol-processing machine like a computer can never be properly described as having a "mind" or "understanding", regardless of how intelligently it may behave. The poor guy in the Chinese room can translate perfectly following the set of rules, but he does not understand a word of what he translates.
Therefore the mind-as-person metaphor is not very helpful to understand how the mind works. Among the many metaphors for the mind, it is probably the least helpful if we want to explain and describe how the mind works. Many metaphors have been invented over time to do this. George Lakoff said:
- "When you start to study the brain and body scientifically, you inevitably wind up using metaphors. Metaphors for the mind, as you say, have evolved over time -- from machines to switchboards to computers." George Lakoff
Curiously the brain always seems to be one of the most advanced technologies that we humans currently have, as Rodney Brooks argues:
- "If we look back over recent centuries we will see the brain described as a hydrodynamic machine, clockwork, and as a steam engine. When I was a child in the 1950's I read that the human brain was a telephone switching network. Later it became a digital computer, and then a massively parallel digital computer. A few years ago someone put up their hand after a talk I had given at the University of Utah and asked a question I had been waiting for for a couple of years: 'Isn't the human brain just like the world wide web?' The brain always seems to be one of the most advanced technologies that we humans currently have." - Rodney A. Brooks
In the mind-as-machine metaphor the mind is a device or machine for doing mental work. The psychologist William James compared the mental life to a stream, a sequence of thoughts and impressions that is never the same thing for long. Maybe the mind-as-container metaphor is more useful. In the Theaetetus, Plato compared the human mind to an aviary, a container where the wildlife of human cognition - thoughts and memories - is kept, broadly restrained but locally elusive. And since Marvin Minsky's "Society of Mind" we know that a society of agents - the mind-as-society metaphor - is probably a better model for human intelligence. In his book, he tries to explain how minds are built from mindless stuff, which he names "agents". Steven Pinker says about it in "How the mind works" (on page 144) "The society of mind is a wonderful metaphor" and continues (on page 419) "Mental life often feels like a parliament within". So what happens if we replace the guy in the Chinese room by the population of a whole country or nation?
If an entire society systematically organizes itself to operate just like a brain, with each individual replacing a group of neurons, then the system will act like a real mind, with mental states, consciousness, etc. The problem is that such a simulation must be fast enough. Let us assume it is fast enough for the moment. Different parts of the population may represent different things. Although no individual knows Chinese, the population as a whole can know the meaning.
Steven Pinker asks in 'How the Mind Works' (p.146):
- "What if we took [a brain simulation computer] program and trained a large number of people, say, the population of China, to hold in mind the data and act out the steps? Would there be one gigantic consciousness hovering over China, separate from the consciousness of the billion individuals? If they were implementing the brain state for agonizing pain, would there be some entity that really was in pain, even if every citizen was cheerful and light-hearted?"
Is there a president somewhere ?
- "No one can whistle a symphony. It takes an orchestra to play it." - H.E. Luccock
The essential thing in the "society of mind" are the agents, and not the president. There is no president giving orders to all the others. According to Daniel Dennett and Steven Pinker, "It's a mistake to look for the President in the Oval Office of the brain". There is no president, no central agency which controls the brain, and the agents which represent the self can not be found in the Oval Office. They are just ordinary agents, although maybe a bit more 'famous' than others.
If the society has no president, why have so many human organizations a president, director or conductor? Does an orchestra really need a conductor or not? As long as everyone has detailed notes and instructions, the organization is clear, and a central conductor, director or leader is not really needed. This does not mean that the system can organize itself, it only means that enough laws and rules already exist to organize the system.
The question if an orchestra needs a central conductor is similar to the question if a company needs a CEO, a country a president or a church a bishop. In organizations, countries or companies, leadership is needed especially if the company or country has just been created (to define a direction) or needs organizational change (to redefine the direction and organization). In the meantime, the CEO or president is often superfluous.
The same argument applies to an orchestra. A conductor is certainly need at the beginning, to synchronize the start, and at the end, to synchronize the ending, and perhaps in the middle to help those who have long breaks (during "organizational change"). All the rest is basically a "show", because all musicians have detailed notes, even for the breaks. The conductor is maybe needed before the actual performance, in order to rehearse the composition, to define the direction: to set the general tempo and speed, to modulate the volume, and to evaluate the overall performance.
Are there any friends, families or social networks ?
Yes, there are. And they are important. Daniel C. Dennett writes in his book "Brainchildren - Essays on Designing Minds", (Penguin Press Science, 1998)
- "In ordinary runs of affairs, large families of beliefs travel together in our mental lives. At one instant, Mary believes that her purse in on the bed, and believes her handbag is on some horizontal surface, and believes that the item containing her comb is supported by the article of furniture one sleeps in, and so forth. Now do all (or many) of these distinct states have to light up and team up to cause Mary to run upstairs? Or is there just one each from the belief family and the desire family that are chosen to do the work?"
The 'society of mind' metaphor becomes interesting if we consider the social networks inside the society. The value of the human mind lies in its vast network which connects the individual agents. To illustrate this point, Marvin Minsky compares in section 4.3 "The Soul" of his book "The Society of Mind" the mind to a painting:
- "The art of a great painting is not in any one idea, nor in a multitude of separate tricks for placing all those pigment spots, but in the great network of relationships among its parts. Similarly, the agents, raw, that make our minds are by themselves as valueless as aimless, scattered daubs of paint."
Relationships exist for example between agents which represent similar objects or situations, or between agents of the same function. According to the BDI model one can identify a belief family, a desire family and an intention family, as Dennett does. Friendships exist if there are any analogies and metaphors (see below).
The social network as a whole is of fundamental importance, because it offers a solution to the hard problem of consciousness: the social network of the mind determines the kind of subjective experience. Everyone has a uniqe “society of mind”, and a unique social network. Each of us is adapted to a slightly different world (or different “slice” of the same world). Already William James said
- "The peculiarity of our experiences, that they not only are, but are known, which their ’conscious’ quality is invoked to explain, is better explained by their relations – these relations themselves being experiences – to one another."
Who decides which agent is active?
First the agents themselves, each active agent activates certain agents in turn, and second the critics in the "jury". Minsky illustrates and describes this idea in his book "The Emotion Machine", where he proposes a coarse “critic-selector” model of the mind, where a number of critics evaluate situations to activate certain selectors, which in turn activate or deactivate clouds of agents. The modulation idea is not bad, although it is not very new or original. Yet critics are a good cue: imagine for example a cartesian theater or a Global Workspace proposed by Baars, where not a single self is sitting in the audience, rather the representatives of the body, like the members of a jury (the physiological counterpart maybe the limbic system and the brainstem). On stage there are agents which represent the current state of the world. The critics in the jury now evaluate every situation, and are able to choose between thumbs up and thumbs down. If they say “Yes, I want to see it again”, then the agents on the stage are likely to appear again, if they say “No, I don’t want to see it”. A nice picture: if a person falls in love, then the critics for love always want to see the same persons on the stage.
What happens when agents disagree ?
The basic idea is simple. Agreement is good, disagreement is bad. Agreement among the agents in the society of mind is associated with pleasure. Pleasure is a state of physical integrity and effortless action readiness of the body. The body signals that it is ready for work, that there are no obstacles or objections for action. It is reflected by an internally consistent representation (no contradictions, consonance, agreement). On the contrary, pain is a state characterized by violation of the physical integrity of the body. The body signals that some actions should be avoided, or that it is tired and depleted. This state is reflected by an internally inconsistent representation (contradictions, dissonance, disagreement).
This model means that physical integrity of the body is connected to the "mental integrity" of the mind. In general, conflicts between agents are bad, while consensus is good. Conflicts represent a situation of cognitive dissonance. They are unpleasant for the mind, society or population, because it means tiresome inhibition or deactivation of agents. Contradictions between agents, for example an incongruity in meaning, lead to confusion and displeasure. The mind must activate more agents, spending more energy to resolve the conflicts. It has a negative return on investment (ROI). The mind makes a large investment to understand a situation and gets little or nothing in return: obviously an unpleasant situation. Displeasure in turn induces a change in behavior: the mind takes steps to prevent more events like this.
To put it in more abstract terms, displeasure arises from a drain for the neural information flow: activity is decreased, agents are deactivated or lost. In the terms of nonlinear dynamics, graph and chaos theory, pain is a sink for neural flow, where agents disappear. Pain, the central mechanism of Self-Protection, is caused by the decay of the system, by an inevitable and unavoidable civil war among agents where agents are inevitably suppressed and the flow of information gets lost. Some agents continuously fight other agents. While displeasure is merely a contradiction between agents, pain can be compared by a heavy struggle or fight between the individuals.
Pain enhances the thoughts of the pain-triggering body-part, but also suppresses strongly all actions involving these body-parts. Some agents are inevitably suppressed, although they are constantly activated. Therefore it creates a 'Sink' for the neural flow, where neural flow is absorbed. Neural flow is the flow of information between the "Mind-agents". In general we find sinks for neural flow unpleasant, and sources enjoyable.
Why is a sink unpleasant? A sink for neural flow is unpleasant, because it decreases the ability of the body to act. It is associated with heavy tasks, which require a large effort, and occurs for example if the body makes actions harder by holding off rewards.
What happens when agents agree ?
Pleasure is associated with a 'Source' of neural information flow, for example the unexpected perception of a desired object which triggers an action to enhance the perception of the object. This action will in turn increase the strength of the pattern for the desired object, which will trigger even more action to enhance it, etc.
If agents activate, support and acknowledge themselves, the mind can deactivate agents, spending less energy to resolve conflicts. It has a positive return on investment (ROI).
Consensus and consistency is also pleasant for the mind, because it means effortless activation of agents. Mutual acknowledgment and confirmation lead to pleasure. Pleasure in turn lets the mind take steps to have more events like this, because they usually involve things that are good for the body (survival) or the race (reproduction). In the worst case, this situation can lead to addiction.
A source for neural flow is pleasant, because it increases the ability of the body to act. It is associated with easy, effortless tasks and occurs if the body has offered a reward. It means that a event has occurred which has been recognized by the body as positive, either in the environment or in the body itself.
A lasting form of agreement can be characterized as 'friendship'. Friendship is a co-operative and supportive behavior between two or more agents. In the society of mind, metaphors and analogies are related to friendships. The process where agents become 'friends' is related to insights and understanding.
What happens when agents change their minds ?
- "The power of intelligence stems from our vast diversity, not from any single, perfect principle. [...] Eventually, very few of our actions and decisions come to depend on any single mechanism. Instead, they emerge from conflicts and negotiations among societies of processes that constantly challenge one another." --The Society of Mind, Chapter 30.8, page 308.
The most interesting things happen when agents first disagree for a long time, then suddenly change their minds and agree with each other. The central element of the situation is the incongruity at the beginning, which is resolved later during the process of insight and understanding.
As said above, an incongruity can be modeled as a dispute or discussion between agents, where agents or coalitions of agents contradict each other. Such a form of incongruity is an essential condition for humor, too. Humor involves the recognition and resolution of an incongruity, and it is closely related to the process of insight itself.
Usually an input from the environment can be represented by different concepts and assemblies, and it can be interpreted in different ways. Which assembly becomes active depends on context and expectations: we expect people to act normal and in the obvious way (not stupid, dumb or crazy), and we also expect that things behave normal and have their usual size. The 'common sense' agent 'task group' acts as a censor which suppresses other representations, even if they are a better match and have no incongruity.
During the punchline in jokes or at the moment of the insight, the censor is suddenly removed or bypassed, and the incongruity is resolved. Sub-groups or coalitions of previously contradicting agents will suddenly reinforce and confirm each other. The unexpected mutual confirmation between agents turns a sink of neural flow into a source. Instead of investing energy to resolve the situation, the mind can hold back energy. A sudden change from negative ROI to positive ROI values occurs, which is evaluated as extremely pleasant. This is what we know as humor.
Fast avalanches of agent activity, which correspond to an exponential growth in activity, are very pleasant - they are related to laughter, humor and amusement. They can be so pleasant, that the activity as a whole must be interrupted by small periodic time periods to avoid an overload - the interrupts we know from laughter. The pleasure we gain from insights is a mechanism to recognize sense in non-sense and the familiar in the unfamiliar.
Our brain is designed to to find meaning in confusion and to solve new problems. It seeks pleasure and happiness, and insights are pleasant while confusion is not.
What happens when agents become aware of themselves ?
If the population is large and complex enough, can it represent and recognize itself? And what happens when agents become aware of themselves ? An interesting question. See collective consciousness.
Consciousness of the self or self-awareness is like a whirl or turbulence in the neural information flow: it is characterized by heavy discussions with few results among *all* agents. Self-consciousness is like the feeling of being strong because you have climbed a huge mountain, but it also like watching yourself from a distance standing on this mountain peak at the same time. It is to recognize yourself (that means every agent is affected) as a small part of the environment (that means only a few special agents are affected).
Moreover, the results of the "discussions" which characterize consciousness on the agent level often interfere with the emotion agents which act through reinforcement. The capacity to act is constrained. The agents are confused and the system eventually shows chaotic behavior. Consciousness is complex. It is the recognition of the familiar in the unfamiliar. Although it is based on the insight of the own existence, it leads to the recognition of the own death. Insight causes pleasure, death the strongest displeasure. So consciousness is related to pleasure in displeasure.
Because the topic concerns the whole system, all agents think it is important. But the insight of consciousness says in turn the system as a whole is only a part of a much larger system. Consciousness is a reconciliation of the whole with the part, similar to eternity in a moment (Goethe's Faust), or a lifetime in a day (James Joyce' Ullysses). A continuous merging (agreement) and splitting (disagreement) of agents. A self-conscious agent is able to see eternity in a moment and a lifetime in a day.
Terms from sociology or crowd psychology
Can we apply terms from sociology or crowd psychology to the society of mind? Terms in question are social movement, collective consciousness, Zeitgeist ("the spirit of the age"), or Historicism (a theory that social events are determined by history and historical conditions). Other interesting candidates are the structure/agency debate or the integration of macrosociological and microsociological perspectives.
Historicism is certainly useful to explain the actions of a society. Hindsight is easier than foresight, we can explain the action of a society or person if we know the history and the historical conditions, but it is very hard to predict future actions. This is true esp. in the case of a criminal: afterwards we can nearly always understand and explain the behavior which has lead to a crime, but in advance it is nearly impossible to predict what someone will do (the stronger the free will we attribute to him, the more difficult it will be).
Links
see the Wikipedia entries for the
- Guardian article about metaphors for the mind What's on your mind?
- David M. West and Larry E. Travis: From Society to Landscape: Alternative Metaphors for AI, AI Magazine Vol. 12 No. 2 (1991) 69-83
What web designers use as a dummy text ("Lorem ipsum dolor sit amet..") to fill empty spaces is in fact an interesting text from Marcus Tullius Cicero. Cicero talks about the conceptions of the most good and most evil, and how they are related to pleasure and pain which are the ultimate motives of desire and of avoidance:
- Marcus Tullius Cicero, De Finibus Bonorum et Malorum
