The Costs of Cognition from an Information-Theoretic perspective: experimental validation – CoCogIT

The proposed projects aim at addressing these questions by taking two approaches:
1. The first approach focuses on studying effort costs and consists in varying cognitive demand (formalized in terms of information theoretic quantities) across blocks and in assessing its impact on behavior and brain activation magnitude. This approach will be used to tackle objectives 1 and 3 and will be implemented in WP1.
2. The second approach of the present proposal focuses more on the consequences of prolonged cognitive activity. It consists in having participants perform tasks of different difficulty/information cost across different sessions and comparing the consequence on behavior, task-related and resting brain activity. The assessment of resting baseline brain activity will rely on Arterial Spin Labelling (ASL), which provides quantitative measures of cerebral blood flow, as well as on BOLD imaging.

Work package 1 – efficient coding of sensorimotor processing in relation fo effort
We decided to develop further the computational framework on which the project was built, by setting up a method for estimating the rate of information transfer during a tracking task, and by dissociating the predictive part (anticipation of the movements of the target ; feedforward processing) from the reactive part (error correction ; feedback processing). A publication is currently underway on this theoretical work.
Moreover, we have proposed a novel theory linking variations of pupil diameter, signatures of arousal changes, to the quantity of information being processed. This theory, published in Proceedings of the Royal Society, allows us to clarify the meaning of this measure, largely used as an index of mental effort.
We have also performed experiments testing our hypothesis of a link between mental effort and information gain in the context of arithmetic tasks. The results, replicated in three successive experiments, have largely confirmed our hypotheses, showing a link between pupil dilation and the quantity of information necessary to update the internal model of the task.

Work package 2 – neural fatigue
We have successfully completed WP2.1 and confirmed our hypothesis. The mere repetition of passive visual stimulation led to an alteration of performance specific to the part of the visual field that was stimulated and was accompanied with a modification of the electrophysiological signal reflecting sensory information processing. The publication of this work is underway.

Neuroimaging data acquisition on mental effort (WP1) will start around 2020. The study on the impact of dual tasks on information processing during ocular tracking (WP1.2) should begin around spring 2021. Moreover, we are preparing two complementary experiments, with a measure of the metabolic rate of oxygen instead of the BOLD signal. This physiological measure should provide a better signature of cortical metabolic activity and we predict thus a more robust relationship between this measure and information rate than the one observed by means of BOLD signal.
However, this technique is not available in Bordeaux, and we will have to conduct these experiments in Brussles, in collaboration with the COSY lab.

We have started acquiring data aiming at measuring the impact of neural fatigue on cortical metabolism, focusing first on the BOLD signal. We have used MultiVariate Pattern Analysis (MVPA) to evaluate the quality of visual stimulus encoding before and after saturation of a part of the visual field through repeated passive stimulation. A collaboration has started with the lab of Kazuhisa Shibata (Riken) to evaluate the impact of neural fatigue on the concentration of metabolites in cortex, via magnetic resonance spectroscopy (MRS). These experiments were initially planned in June 2020 but have been reported to November 2020 because of the sanitary situation. Another collaboration with the COSY laboratory from UCL, Brussels, will allow us to evaluate the impact of neural fatigue on resting metabolism in cortex by means of ASL. These experiments should start in Brussels in the beginning of the year 2021.

Article
Zénon, A., 2019. Eye pupil signals information gain. Proceedings of the Royal Society B 286(1911). royalsocietypublishing.org/doi/pdf/10.1098/rspb.2019.1593

Communications
1. “Mental effort, pupil dilation and information gain.” Geneva Neuroscience and Biotech Campus, Oct. 2019
2. “Cognitive correlate of pupil dilation” Gent, Faculty of Psychology, Février 2019.

The origin of the aversiveness of intense cognitive activity and the mechanisms that lead prolonged mental activity to induce fatigue remain neuroscientific enigmas. The objectives of the present proposal consist in providing experimental validation to a novel framework to understand the origin of cognitive cost based on information theory and which predicts that subjective effort costs and task aversiveness are a function of the information cost of the task. We propose three hypotheses regarding the nature of this relationship, corresponding to three different mediating factors: local capacity limitation, global capacity limitation and metabolic alteration rate. We will test these hypotheses experimentally, by means of behavioural and neuroimaging (BOLD and Arterial Spin Labelling) techniques in humans.