GATSBY COMPUTATIONAL NEUROSCIENCE UNIT
UCL Logo

The Freedom of Information

Simon Laughlin and Jeremy Niven

Department of Zoology, University of Cambridge, UK

Cognition, and more generally adaptive behaviour, depends upon an organism’s ability to specify its state, and the state of its environment, past, present and future (e.g. the distribution of sensory stimuli, the distribution of muscular activity and the patterns of synaptic connectivity that depend upon experience). By definition, signals that specify states carry information. Consequently, despite its seemingly abstract qualities, Information Theory is, and will be continue to be, a useful tool for understanding the evolution and generation of cognition.

Causal relationships generate information because by determining the state of the effect you decrease uncertainty about the cause. It follows that, as David Dusenbury (1996) said “Information is where you find it”. This insight is born out by the multiplicity of sensory cues that confound studies of animal behaviour, and the shocking promiscuity of signaling in the brain.

But, although information is freely available, it does not come for free. The acquisition, processing and delivery of information require that the necessary states must be determined and coupled. This boils down to narrowing down the origins, compositions, magnitudes and destinations of signals and this process of confinement (generating information by restricting the freedom of signals) requires resources; energy for new and higher quality signals, and space and materials for additional pathways and processes.

We will review work (Niven and Laughlin, 2008), principally on visual systems, which analyses the material and energetic costs of confining and specifying signals so that the information that enables appropriate behavioural responses and supports cognition can be extracted and transmitted. A number of studies show that some neural systems have evolved to operate efficiently within constraints imposed by the basic biophysics of generating and connecting informative signals and provide evidence that the costs of information are balanced against its benefits.

 

BACK