Collective Intelligence

A collective phenomenon is more the product of, rather than the sum of, its parts, and has been explained by Cal Tech biophysicist John Hopfield (1982) whose "neural net" models are collective.

Suppose you put two molecules in a box, every once in a while they collide and that's an exciting event. ... If we'd put ten or even a thousand more molecules in the box all we'd get is more collisions. But if we put a billion billion molecules in the box, there is a new phenomenon-sound waves.

Figure 1.3: Axoplasmic transport occurs by coordinated activities of microtubule attached sidearm, contractile proteins ("dynein") which cooperatively pass material in a "bucket brigade." The orchestration mechanism is unknown, but shown here as the consequence of signaling by "soliton" waves of microtubule subunit conformational states. By Fred Anderson.

Figure 1.3: Axoplasmic transport occurs by coordinated activities of microtubule attached sidearm, contractile proteins ("dynein") which cooperatively pass material in a "bucket brigade." The orchestration mechanism is unknown, but shown here as the consequence of signaling by "soliton" waves of microtubule subunit conformational states. By Fred Anderson.

Observation of two, or ten, or thousands of those molecules would not suggest the Mozart or Madonna that can arise in a collection of more than a trillion trillion molecules. Other examples of collective phenomena may be seen in beehives, ant colonies, football teams, governments and various types of material phase transitions. For example, superconductivity and magnetism are collective effects which occur in certain metals as their individual atoms come into alignment. By cooling these metals, thermal fluctuations cease, atoms become highly aligned, and below a critical temperature totally different qualitative properties of superconductivity or magnetism emerge.

How might collective phenomena be tied to consciousness? Brain neuron synaptic transmissions are relatively slow at several milliseconds per computation-they are about 100,000 times slower than a typical computer switch. Nevertheless vision and language problems can be solved in a few hundred milliseconds or what would appear to be about 100 serial steps. Artificial Intelligence (AI) researchers conclude that this computational richness is accounted for by collective effects of parallelism and rich interconnectedness. With billions of neurons, and with each neuron connected to up to hundreds of thousands of other neurons, Al "connectionists" view the brain as a collective phenomenon of individually stupid neurons. Groups of highly connected neurons are thought to attain intelligent behavior through properties of feedback and reverberation. Walter Freeman (1972, 1975, 1983) of the University of California at Berkeley contends that a "critical mass" of about 100,000 neurons yields intelligent behavior. However, intelligent behavior occurs within nematode worms of 1000 cells and 300 neurons, within cytoplasm in single cell organisms and within single neurons. Individual neurons with tens to hundreds of thousands of connections cannot be stupid and fulfill their multiple functions, integrate input/output and modulate synaptic connection strength. Each nerve cell is a sophisticated information processing system in and of itself! The cytoskeleton within neurons and all living cells is a parallel connected network which can utilize its own collective phenomena to organize and process subcellular information (Figure 1.3). The cytoskeleton can convey analog patterns which may be connected symbols (Chapter 8). Although overlooked by AI researchers, the cytoskeleton may take advantage of the same attributes used to describe neural level networks. Properties of networks which can lead to collective effects among both neurons and cytoskeletal subunits include parallelism, connectionism, and coherent cooperativity.

0 0

Post a comment