This proposal aims at defining an area of collaborative research between neurobiologists, computer scientists and mathematicians, which will investigate the role that rhythmic neural activity may play in the brain and in a brain-inspired computational architecture for artificial cognitive systems of the future.

Rhythmic activity is prevalent in many regions in the brain and has been extensively investigated by neurobiologists, both experimentally and using computational models. Specific rhythms, referred to according to their respective frequency ranges as “theta” and “gamma” rhythms, have been particularly implicated in both perceptual and memory mechanisms and systems in the brain, both working and long-term episodic memory.

Any artificial cognitive system which employs a brain-like architecture in which representations of perceptual and other events are not given an explicit “time-stamp” must employ a mechanism which will allow a correct temporal organisation of these events to be represented for both memorisation and recall.

The hypothesis is that rhythmic activity may play a fundamental role in this respect within the brain and may thus be a crucial mechanism to replicate in a brain-inspired architecture for artificial cognitive systems. In addition, a greater systematic understanding to the role of rhythmic neural activity in the brain will lead to a better understanding of how deficiencies in human perceptual and memory abilities, in particular working and episodic memory deficits, might be related to failures in the neural mechanisms for the generation and control of rhythmic activity.

Rhythmic activity is ubiquitous in many areas of the invertebrate brain, ranging from slow <1Hz oscillations which are found in thalamus during sleep, through the 5-7Hz “theta” rhythm, predominantly in medial temporal lobe structures, which is observed in alert and active animals, and has been linked to the operation of memory systems, to higher frequency neural activity, in particular the 30-80Hz “gamma” activity which is prevalent throughout the sensory areas of the brain, and in medial temporal areas linked to memory, and has been proposed as playing a role in the problem of perceptual “binding”.

In particular the theta and gamma oscillations found in hippocampus and elsewhere and the neural mechanisms which support these rhythmic activities have been extensively investigated by both neurobiologists and scientists currently working at the interface between computer science and neuroscience (eg [1]-[3]). However the precise cognitive role that such rhythmic activity plays in perception, working and long term memory have yet to be fully elucidated.

Future artificial cognitive systems which use computational architectures based on neurobiological principles of information processing will employ, as in the brain, vast numbers of processors which will be interconnected both by local circuits into neural assemblies which are adaptively and asynchronously formed in order to carry out sharply defined perceptual, cognitive or motor functions, and into more wide spread systems, involving many brain regions, for supporting memory, attention, planning, decision-making and other higher level cognitive processes.

A significant problem, in such a computational architecture, is that of time, in particular the representation of temporal order in such neural processing circuits and systems. It is not at all clear as to how this problem might be explicitly dealt with. The fact that perceptual representations are ordered temporally is vital for working memory tasks, eg remembering a sequence of numbers to make a ‘phone call, and in long term episodic memory, in which personally experienced events are predominantly recalled as temporally ordered sequences of perceptual experiences.

It is unlikely that the brain functions in a manner, which allows individual experiences to be explicitly “time-stamped”, as they might be in a conventional computer architecture, which employs a real-time clock. So there must be mechanisms in a brain-inspired computational architecture, as in the brain, which organise the recall of memorised perceptual experiences in the correct order. The hypothesis that will be addressed in this “grand challenge” is what mechanisms can play such a role, and in particular whether rhythmic neural activity is the key mechanism involved.

Whilst there has been a great deal of detailed investigation of the mechanisms and properties of rhythmic activity in the brain, there remains a major challenge in understanding its role at a system level, in perceptual and memory (working and long-term episodic) systems.

There is substantial opportunity for a collaborative activity between neurobiologists and computer scientists to engage in the investigation of this major feature of brain activity, in order to establish whether it might be a major component of future brain-inspired computational architectures for artificial cognitive systems. This activity is envisaged, and will be initiated, as a collaboration between neurobiology, computational modelling, computer architectures, and mathematics.

How does the brain encode time?