Community Research and Development Information Service - CORDIS


BIOMOTIV Report Summary

Project ID: 260747
Funded under: FP7-IDEAS-ERC
Country: France

Final Report Summary - BIOMOTIV (Why do we do what we do? Biological, psychological and computational bases of motivation)

Why do we do what we do? We are largely unaware of the mechanisms that determine our behavior: how we form goals and how these goals translate into action. A goal can be reduced to an expected reward, i.e. an anticipated situation with a positive value: earning money, winning awards, being loved, etc. Uncovering how the brain assigns subjective values to potential situations is central in this research project, which seeks to understand: a) how contextual factors (e.g., background music or social situation) affect neural representation of value, b) how these neural values determine behavioral outputs (i.e., likeability rating, decision making, motor or cognitive performance) and c) how these neural values are formed and updated though associative learning and social interaction.

To answer these questions, our team combines three approaches: 1) human cognitive neuroscience, which is central as we ultimately wish to understand ourselves, as well as human pathological conditions where motivation is either deficient (apathy) or out of control (impulsivity), 2) primate neurophysiology, which is essential to describe information processing at the single-unit level and to derive causality by observing behavioral consequences of brain manipulations, 3) computational modeling, which is mandatory to quantitatively link the different description levels (single-unit recordings, local field potentials, hemodynamic signal, vegetative manifestations and motor outputs).

So far, we have delineated the brain network that represents both positive values (expected rewards) and negative values (expected punishments). We have established quantitative links between value signals encoded in the electrophysiological activity of neurons recorded in monkeys, and those encoded in the hemodynamic activity of brain regions recorded in humans. Next we have identified several fundamental properties of the neural code for subjective values: the brain valuation system is generic (signals values for various categories of items), automatic (signals values even when the brain is engaged in a distractive task), aggregative (mixes up values of unrelated contextual features), and auto-correlated (value signals depend on previous activity). We have also characterized interactions of the brain valuation system with other brain systems (such as the perceptual, motor, cognitive, episodic and mirror systems) that can impact on, or be impacted by, subjective values. These neural interactions provide explanations for a number of psychological phenomena, such as choice impulsivity (why we often succumb to the attraction of immediate pleasures) or mimetic desires (why we often pursue the same goals as other people).

Finally, we have built a computational model of motivation, formalized as a function that adjusts the direction and intensity of behavior according to some arbitration between expected costs and benefits. This model can be fitted to the behavior observed in a battery of motivation tests that we have now finalized. These tests involve key motivational factors (reward, punishment, effort, delay, etc.) and processes (judgment, choice, performance, learning). To analyze the behavior in these tests we have developed novel Bayesian methods that enable inferring model parameters and comparing variants of the model. Fitted parameters characterize motivational dispositions such as sensitivity to reward, punishment, effort, delay, etc. Thus model-based analyses of the behavior provide computational phenotypes that may characterize a clinical state. We already obtained proofs of concepts for this approach to give insight into motivation deficits in neurological and psychiatric diseases, as well as motivational effects of classical medication targeting neuromodulatory systems (dopamine, noradrenalin and serotonin), in both human and non-human primates.

Our long-term objective is to build a comprehensive neuro-computational model that would account for the psycho-physiological determinants of human behavior. This is important for the clinics, as no such theory exists yet that could inform the design of diagnostic and therapeutic tools for motivational syndromes manifested in neuropsychiatric disorders. This is also important for economics, as this theory would explain a number of deviations to rationality that are linked to the heuristics developed by the primate brain through natural selection.

Reported by

Follow us on: RSS Facebook Twitter YouTube Managed by the EU Publications Office Top