Chaos in the mind
This page is about the role of chaos in neural networks at the interface between neuroscience and psychology. One major role may be to guarantee creativity and diversity of behavior. The unconscious mind takes the flood of information and simplifies, canalizes and categorizes it into ever-changing streams of patterns and packages. Chaos arising from combinatorial search maybe important as a mechanism to control the steam of consciousness (at least the unconscious part), to arrive at novel solutions for cognitive problems, to find a way or trajectory to the best idea which matches the current situation, and to find new analogies, metaphors and insights. A number of scientists have coined names for this process, Peter Ashwin's has proposed "Cycling Chaos", Mikhail Rabinovich a "Winnerless Competition" (WLC), and Ichiro Tsuda "Chaotic Itinerancy". They are all similar, the general idea is to connect different saddle points with "cycling" trajectories.
Chaos and the point of view
Neural networks, the building blocks of a neural assemblies, show inherently complex and chaotic behavior. The more detailed you model them, the more complex the situation becomes: if you simulate a single neuron exactly with differential equations, even a single neuron can show complex (for instance spiking and bursting) or even chaotic behavior. The differential equations for single neurons can show chaotic behavior (Hodgkin-Huxley model, Fitz-Hugh-Nagumo model,..). Bursting of neurons can be described by bifurcation theory. Neural networks are inherently complex because every connection involves a merging and splitting operation. When the brain processes information, in every step the previous state is merged and splitted in complex ways. It we consider the overall network, for example with an EEG, then we see that EEG data apparently is chaotic.
Yet on the psychological level, chaos seems to play a minor role. For most of our actions we can recognize a certain cause and identify an emotion, a motive or motivation, which has triggered this action. If you go from the bottom (neuroscience) to the top, everything seems to be complex, chaotic and involved in chaos, from the attempt to describe even a single neuron to a typical EEG. If you look from top (psychology) to the bottom, there seems to be little chaos. Somewhere in between chaos seems to vanish, and maybe it is in fact the essential connection, or at least involved in it.
It all depends on your point of view. It's like the "Free Will" problem. If you think of your brain as a large deterministic neural network (the bottom-top view of neuroscience), everything is determined. Although the behavior of the network maybe is chaotic, in principle you can calculate the action from the perception and there is no free choice. If you think of your brain as the seat of your personality, your identity and your soul (the top-bottom view of psychology), the soul has the free choice to select an action : there is nobody else who can influence the thought processes in your head directly, only yourself. No one can force or coerce you to think something.
It is like the difference between past and future. In the past everything seems to be determined, nothing is changeable, and for most of the human actions you can give a good explanation. In the future nothing seems to be determined, everything is changeable, and you can not predict human action. The interface in the stream of time, the short time where the future become the past, is the place where things are toppled, where the undetermined things became determined, and the changeable became unchangeable. (see Paul Watzlawick in "How Real Is Real?: Confusion, Disinformation, Communication" he also gives a beautiful quote of James Joyce from "Finnegans Wake". James Joyce asks, if history is like a mighty stream, which can be overseen by the gods from the infinite past to the infinite future, if the Free Will is only an illusion, while the stream of existence leads us in an already existing future, if history is an already finished film, which is projected on the screen of the present time.)
Free will is the possibility to choose, it is the free future choice you seem to have. The past choice which has been chosen is seldom free. The deliberate thinking self is squeezed between social constraints (norms, laws and obligations) at the top and emotional constraints (feelings,drives and needs) at the bottom. We do have a free-will if we take the "top-bottom" view of psychology and consider future actions. The future choice is free and undetermined. We don't have a free-will if we take the "bottom-up" view of neuroscience and examine past actions. The real and past choice is determined, fixed and not free from any influence.
So the most interesting things happen in the middle between neuroscience and psychology, where determined things became undetermined and vice versa. They can not be found on the level of individual neurons, nor on the top level of total behavior. Donald O. Hebb (1904-1985) has recognized this, and in his masterpiece "The Organization of Behavior" he tried to connect both sides, the neural and the psychological side. He identified the assembly as the basic element and the building block, the interface between bottom and top level behavior.
Chaos and recombination
What is the role of chaos in information processing ? Perhaps the comparison with information processing in other systems, for example genetic systems, can give us a hint. Through DNA recombination, information is stretched (between two DNA sequences, different kinds of combinations - each of them different from the parent sequences - are possible) and folded (two DNA sequences are combined to only one sequence). Repeated DNA Recombination produces a wide variety of DNA types. The room of all possible sequences is covered by chaotic search, until an optimal sequence is found. This "optimal" sequence is the one with the best fitness, which is optimal adapted to the environment.
Stretching and Folding (Variation and Combination, Divergence and Convergence) of information results in increased complexity and diversity, in dynamical systems the behavior can be chaotic. Chaos that can be useful for a chaotic search of an optimal DNA sequence - or an optimal combination of assemblies.
Can we compare the recombination of neural assemblies with the recombination of DNA ? Perhaps the same mechanism can be applied to neural networks and certain kinds of neural assemblies ? Consider for example a neural network, where some method of inhibition allows only a certain amount of assemblies to be active at the same time. An assembly would receive informations from many assemblies (merging) and would send information to other assemblies (splitting). Similar to recombination of DNA, where an optimal combination is found by chaotic search, the recombination of neural information could lead to an optimal set of assemblies, which matches the input perfectly. Different neural assemblies could excite other neural assemblies, which in turn could activate other assemblies... A recombination of assemblies would take place until one combination is found which is optimal for the current input or context. This mechanism would work below the surface of consciousness, if the result, the final combination of assemblies, is the real "thought". If it is involved in the process of insight itself.
The idea of "Recombination of Neural Assemblies" is not new. It can be found in the work of Gerald Edelman, too. Gerald Edelman won a Nobel Prize in 1972 for establishing that the immune system works according to Darwinian principles. He also tried to apply this principles to the Brain in his Books "Neural Darwinism, The Theory of Neuronal Group Selection" "Bright Air, Brilliant Fire: On the Matter of the Mind " Terrence Sejnowski from the Salk Institue says they are turgidly written. And he is right, they are hard to read. But Edelman is not anyone, he is the director of "The Neurosciences Institute" in La Jolla, California. He is probably not completely on the wrong track.
Role and Function of Chaos
- "Chaos underlies the ability of the brain to respond flexibly to the outside world and to generate novel activity patterns, including those that are experienced as fresh ideas." (Walter J. Freeman, The Physiology of Perception)
The mechanism described above uses chaotic search to find the optimal combination of neural assemblies. Chaos helps exploring possibilities. If you ask "what is the function of chaos/randomness" this is definitely one answer. Chaotic search is deterministic (which means that the system has no random or noisy parameters) and yet irregular and unpredictable like random search. When a system is chaotic, you can only predict it's future in a probabilistic manner. A system undergoing chaotic interactions samples all possible states available to it, it is running through all possible states.
In the case of a simple system (for example the Lorenz attractor), normally you explore the parameter space of the system to find the particular values with the most interesting chaotic behavior. But chaotic behavior can be used to explore parameter spaces as well, for instance to find the parameter configuration with the highest fitness in a complex adaptive system.
Playing means to sample possible situations, too. Why do children play ? Because it's fun, in order to explore and understand their environment. A child's job is to play, and most of a child's developmental needs are met through his/her desire to play, for example learning hand-eye coordination, understanding cause and effect, participating in society, dealing with conflicts,..
Has chaos something to do with playing ? Childern play because it is fun to get new insights. Chaos helps to get new insights and to increase flexibility. A function of chaos is to enable creative thinking and to help to create new insights. Perhaps the process of perception (getting an "insight") can be described as the transition from one stable state to another, or the creation of a new cognitive rule or connection, which leads to a new stable state (of lower "Energy", since "Energy" is released during the transition). A stable state is made of several assemblies, which enhance and excite each other. It is not static, but includes dynamic and chaotic fluctuations. This chaotic fluctuations help to make new transitions and enable creative thinking.
So the role of chaos maybe is to increase intelligence (the diversity and complexity of behavior, and the ability to increase them if necessary), because it seems to be involved in exploration, creation of new ideas and insights.
- Lewis Dartnell, Chaos in the Brain