Thursday, November 1, 2007

LolAntz


In my last posts I wrote about the question how AI and Robotics can tell us something about the architecture of (simple) biological systems. In this post I’ll give an example of how the study of simple organisms, augmented by AI and robotics, can give us information about the structure of cognitive systems.
Studying ant’s navigational skills is such an example. It can tell us something about the evolution of cognitive mechanisms in general because an ant's “domain-specific processing modules” (Wehner 2003a: 585) show signs of modular adaptation, that is, it seems to have
“evolved to solve particular problems encountered by Cataglyphis during its foraging lifetime.” (Wehner 2003a: 579).
In detail, the ant’s navigational toolkit consists of:
“its skylight (polarization) compass, its path integrator, its view-dependent ways of recognizing places and following landmark routes, and its strategies of flexibly interlinking these modes of navigation to generate amazingly rich behavioural outputs.“ (Wehner 2003a: 579)
The sophistication of these synchronically orchestrated navigational modules is quite amazing given that they are found in a brain that weighs 0.1 mg. The human brain weighs about 13 million times at much, about 1,300 g (Jones 2004).
It is probable that our brains too, consist of cognitive mechanisms (and learning mechanisms) which are
"hierarchically nested adaptive specializations, each mechanism constituting a particular solution to a particular problem” (Gallistel 2000).
Another fascinating feature of the ant’s navigational toolkit is its context-dependency . Cataglyphis doesn’t create a ‘mental map’ of its environment, but has a highly egocentric perspective and employs a ‘path integrator’, a permanently updated system informing the ant “about its current position relative to its point of departure”, a little like Ariadne’s thread. (Wehner 2003a).
If the ant is picked up and placed somewhere else, it as a hard time getting home (if it even makes it at all) because the neurological ‘thread’ of the ant’s navigational system isn’t connected to the starting point anymore. When the path-integration vector (the ‘thread’) is displaced, what otherwise would leave the ant straight home is now worthless, because any information about the environment is strictly evaluated in relationship to the ant itself. Therefore the navigational toolkit of Cataglyphis is not only an example of domain-specific adaptation, but also of embodied intelligence (Wehner 2003b).

Interestingly, an experiment by Floreano and Nolfi (1996) in which robots had to explore an area (which I described in my first post) led to the development of very similar mapping-strategies in the robot’s self-organizing neurons. (Wehner and his colleagues also suceeded in building a mobile robot modeling the navigational skills of Cataglypghis, with comparable results.)
This means that both in Cataglyphis and simple artificial systems,
“There is no categorization of the environment that is independent of it” (Poirier et al. 2005: 751)

These “lessons from Cataglyphis” (Wehner 2003b), crucial as they are for understanding human cognition, surely aren't the whole story, especially as there is evidence for a dual-system account of human cognition, consisting of a ‘primitive’, fast , automatic and strongly modular system, and a more-fluid, conscious, cross-domain system. (Evans in press) Additionally,
“One should never underestimate the functional economy of nervous systems: once they have been adapted, over evolutionary time, to the principal physical properties of a predictable environment, they can employ comparatively simple neural strategies to solve quite sophisticated computational tasks.” (Wehner 2003a: 582)
A higher-cognitive task like reading, for example, can probably be explained best by the functional economy and plasticity of human brains. For reading, the “capacity to accommodate a broad range of new functions through learning” (Dehaene 2004) and the ability to employ (or ‘recycle’ as Dehaene calls it) neuronal circuits which serve a similar or related function seems to be essential.
Of course there is still a long way from systems showing properties of embodiment to systems that have internal perspective, but the research discussed in this as well as other posts makes it pretty clear that embodiment is without doubt a crucial component not only of all sensori-motor systems, but also of cognitive systems (Cruse 2003),

P.S.: please note that the ant in the picture isn't a desert ant but a leafcutter ant. I just couldn't find a useful picture of cataglyphis to toy around with ;-)

References:

Cruse, Holk. 2003: “The Evolution of Cognition – A Hypothesis.” Cognitive Science 27: 135–155

Dehaene, Stanislas. 2004. “Evolution of human cortical circuits for reading and arithmetic: The “neuronal recycling” hypothesis." From monkey brain to human brain. Eds. S. Dehaene, J. R. Duhamel, M. Hauser & G. Rizzolatti Cambridge, MA: MIT Press.

Evans, Jonathan St. B.T. in press. “Dual-Processing Accounts of Reasoning, Judgment, and Social Cognition.” Annual Review of Psychology 59

Floreano, Dario, and Francesco Mondada (1996), “Evolution of homing navigation in a real mobile robot”, IEEE Transactions on Systems, Man, and Cybernetics – Part B: Cybernetics 26:396–407.

Gallistel, C.R. 2000. „The Replacement of General-Purpose Learning Models with Adaptively Specialized Learning Modules.” The Cognitive Neurosciences. 2d ed. Ed. M.S. Gazzaniga. Cambridge, MA: MIT Press: 1179-1191.

Jones, Owen D. 2004. “Law, Evolution, and the Brain: Applications and Open Questions.” Proclamations of The Royal Society of London B: Biological Sciences 359: 1697-1707.

Poirier, Pierre, Benoit Hardy-Vallée and Jean-Frédéric Depasquale.2005. “Embodied Categorization.” Handbook of Categorization in Cognitive Science. Eds. Henri Cohen and Claire Lefebvre. Amsterdam: Elsevier, 2005.

Wehner, Rüdiger. 2003a. “Desert ant navigation: how miniature brains solve complex tasks.” Journal of Comparative Physiology A: Neuroethology, Sensory, Neural, and Behavioral Physiology 189: 579–588

Wehner, Rüdiger. 2003b. “Blick ins Cockpit von Cataglyphis” Naturwissenschaftliche Rundschau 56.3: 134-140.

No comments: