Brain Network Dynamics: Brain as Anti-Algorithm
Source: Al Fin
Cognitive scientists are uncovering more secrets of the brain every day. One fascinating line of brain research involves how the brain forms categories and metaphors.
At the IMP in Vienna, neurobiologist Simon Rumpel and his post-doc Brice Bathellier have been able to show that certain properties of neuronal networks in the brain are responsible for the formation of categories. In experiments with mice, the researchers produced an array of sounds and monitored the activity of nerve cell-clusters in the auditory cortex. They found that groups of 50 to 100 neurons displayed only a limited number of different activity-patterns in response to the different sounds.
The scientists then selected two basis sounds that produced different response patterns and constructed linear mixtures from them. When the mixture ratio was varied continuously, the answer was not a continuous change in the activity patters of the nerve cells, but rather an abrupt transition. Such dynamic behavior is reminiscent of the behavior of artificial attractor-networks that have been suggested by computer scientists as a solution to the categorization problem. _SD
Here is the study abstract from Neuron:
The ability to group stimuli into perceptual categories is essential for efficient interaction with the environment. Discrete dynamics that emerge in brain networks are believed to be the neuronal correlate of category formation. Observations of such dynamics have recently been made; however, it is still unresolved if they actually match perceptual categories. Using in vivo two-photon calcium imaging in the auditory cortex of mice, we show that local network activity evoked by sounds is constrained to few response modes. Transitions between response modes are characterized by an abrupt switch, indicating attractor-like, discrete dynamics. Moreover, we show that local cortical responses quantitatively predict discrimination performance and spontaneous categorization of sounds in behaving mice. Our results therefore demonstrate that local nonlinear dynamics in the auditory cortex generate spontaneous sound categories which can be selected for behavioral or perceptual decisions. _Neuron Article Abstract
Here is a broader look at brain network dynamics in the context of decision making:
Cortical network dynamics of perceptual decision-making in the human brain
Brain cells work together in groups, in a dynamic fashion.
Spontaneous rhythmical activity occurs in groups of neurons -- whether artificially cultured in the lab, or in self-selected groups within a living brain.
When separated groups of neurons communicate with each other over a distance in the brain, they utilise a method of synchronous oscillations -- a language that scientists have just begun to understand.
Billions of dollars are spent every year on the quest to achieve human level artificial intelligence. Most of this research is based upon algorithmic design, utilising digital computers. But as anyone can see from looking over recent findings in the neuroscience of cognition, the brain is more of an anti-algorithm. The logic of brain network dynamics has almost nothing in common, conceptually, with the algorithmic basis of digital computing.
AI researchers have attempted to narrow the conceptual gap by utilising "neural net computing," "fuzzy logic computing," and "genetic algorithmic computing," to name three alternative approaches. And these alternative approaches are likely to be very helpful in both applied and theoretical computing and information science. But do they get AI researchers closer to the goal of human-level machine intelligence?
Probably not. Not even the startling potential of memristors and similar semiconductor devices are likely to close that gap appreciably.
As discussed recently in an article quoting quantum physicist David Deutsch, artificial intelligence research is desperately in need of better supporting philosophical structures.
Until then, it is likely that artificial intelligence research will continue to spin its wheels pursuing better algorithms to emulate the brain, without a good understanding of what the brain does.
It is possible to emulate the human brain, using an approach that depends to a limited extent upon algorithmic control, in conjunction with other conceptual methods. But not before researchers learn to approach the problem in entirely new ways, on new logical levels..
| }
|