Ant Decision Making

Standard

The dynamics of physical systems have to follow common principles.  After skimming this paper it seems the brain have similar “decision making” properties.  To wit…

“Ant foraging is a paradigmatic example of self-organized behavior”

“…this process is not necessarily deterministic: probabilistic behavior of individuals play a very important role”

The collective capabilities of social insects are known to be largely the result of a nonlinear cooperative phenomenon.  Typically, individual ants display a limited repertoire of activity patterns.  Unlike individuals, the colony as a whole display very complex patterns, involving search, information transfer and some computational processes.  Examples of such computational abilities are task allocation or the decision between two different food sources.  In this sense, when looking at an ant colony, we perceive several properties in their organization which are commonly shared with the brain and eventually with standard neural networks (NN). These common properties have been listed in several studies and are summarized by:

  • robustness of behavioral patterns against noise
  • collective decision-making
  • emergent computation.

In ant colonies, a parallel distributed processing of information is performed.  [This is also the current best model of brain processing! – “parallel distributed processing of information”]

Individuals gather information from the environment as well as from their nestmates. As a consequence, a given distribution of tasks is always present…much of the structure of the ant colony is based on order parameters, defined as the proportion of individuals existing in one state or another.  {hmmm? Individual ants play a similar role to individual neurons.]

As they say: “the most striking social effects turn out to be the holistic outcomes of mass communication combined with the rise and decline of pheromones and foodstuffs.” There is an important property that makes insect societies rather different from brains: the system is fluid, i.e., information transfer is gathered by moving entities.  The “connection” among individuals is a transient phenomenon, unlike synaptic connections, which are (more or less) fixed.  It is well known that in the last case, a neural network can (under some constraints) perform as an associative memory.  The power of NN is undiscussed, but how far can go, computationally, those dynamical systems sharing some properties with NN but with no fixed connectivity?

information processing in complex systems:
–  To store information, the system has to be able to stabilize the attractors in deep enough minima (i.e., those defined by F (m,h)) .
–  But in order to process information, switching among attractors is necessary.

If, through some self-regulated mechanism, a switch among attractors is available, processing becomes possible. Though it has been suggested that complex computation takes place in systems poised at critical points, here we suggest a different strategy. The system can store information by means of attractors and switch among them by moving through critical points.

“…(local information) is judged principally, and perhaps exclusively, by the “electorate” response of the colony through all-or-none “voting” by the individual ants”

The external signals are appropiately amplified by the ants, and the self-reinforced field acts on the individual ant states through a change in the transition probabilities.  By depending on the density of ants and the rate of decay of the chemical field, the external inputs can be amplified.  Then an emergent pattern is obtained: a self-sustained chemical field is created.  As a consequence a global colony organization is reached.  Some parameter combinations (r, m) makes the system more or less flexible, eventually switching back towards other attractors as the external inputs are removed.  this idea, we have obtained a NOR gate in several ways.  One or two chemicals can be used. They can act on all types of ants or in different ways for different states.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s