Brain mechanisms of dynamic behavioral adaptation – DYNADAPT

DYNADAPT’s hypotheses are based on my recent research: 1) the relationships between FB, AiDE, and behavioral adaptation are built through learning thanks to a network specifically including MCC and DLPFC. 2) The anatomo-functional organization of this network has evolved from monkey to chimpanzee, and to human. Specifically, the chimpanzee shows an organization similar to human and the macaque has all the first fruits of this organization. These data suggest that i) the macaque is a good model to study the causal role of this network on behavioral adaptation and on the global functioning of the brain, and ii) the analysis of the organization of this network in the chimpanzee constitutes the key to bridging the macaque and humans and identifying what makes the human brain unique. To test these hypotheses in human and non-human primates, DYNADAPT uses the following approaches: 1) the network involved in behavioral adaptation in humans is determined using functional Magnetic Resonance Imaging (fMRI) and a new adaptation task allowing the study of learning the relationships between FB, EIA and behavioral adaptation. 2) the homologue of this network in the macaque and the chimpanzee is identified thanks to the coupling between resting fMRI (RS-fMRI) and a new analysis consisting in comparing the «fingerprints« of the connectivity of this network in the 3 species. 3) to determine the causal role of this network on i) overall brain activity and b) adaptive behaviors thanks to the combination of transient pharmacological disturbances in macaques and a) RD-fMRI and b) analysis of behavioral deficits in the adaptive task, respectively

Preliminary results concerns Objectives 1 and 2.
Objective 1. Preliminary results in 8 subjects show that the MCC-DLPFC network is involved in the analysis of FB and AiDE. Importantly, because naïve subjects were involved, our data show how this network learns these various aspects, and reveal the additional involvement of the amygdala in early phases of learning.
Objective 2. The first aim of this objective was to identify the macaque homologues of the human MCC and DLPFC regions involved in adaptive processes in order to determine the critical seeds for pharmacological perturbations in Objectives 3/4. For this purpose, 1) RS-fMRI data have been obtained the 8 human subjects involved in Objective 1 and are currently analyzed, 2) two macaques are currently trained to fixate a central fixation cross in a MRI mock scanner in order to obtain RS-fMRI data in awake macaques. The second aim was to reconstruct the evolution of the adaptive frontal network through the primate order, from macaque, chimpanzee, to human, thanks to the analysis of RS-fMRI data. In addition to the RS-fMRI data in macaque and human that are in acquisition process (see above), RS-fMRI data of 4 anesthetized chimpanzees have been analyzed. In addition to this work initially planned, I also assessed the morphological sulcal organization of the MCC in macaque, baboon, chimpanzee, and human and showed that precursors of the human sulcal organization of the MCC can be found in macaques (Amiez et al. 2019). In addition, this study revealed that an aspect of the human sulcal organization of the MCC (the presence of a paracingulate sulcus) was not human specific as previously thought, but is also present in chimpanzee (Amiez et al. 2019). The discovery that the sulcal organization in the MCC is a good marker for understanding homologies across primates is major and will help the identification of inactivation seeds for pharmacological perturbations planned in Objectives 3 and 4.

DYNADAPT will provide critical insights into how the MCC-DLPFC network is organized to produce optimal adaptive behavior through the primate order to reach its highest level of organization in human. As such, it will provide a breakthrough in heterogeneous domains.
The first impact will occur in basic cognitive neuroscience, with new understanding of the function of a network involved in behavioral adaptation. The second impact will occur in clinical cognitive neuroscience and psychiatry. The MCC-DLPFC network is widely dysfunctional in a large battery of neurological and psychiatric diseases. Highly specific alterations within MCC and DLPFC sub-regions should reveal specific neurophysiological and behavioral alterations (markers). As such, DYNADAPT will provide understanding of the physiopathology within this network and new guidelines for better treatments. The third impact will occur in comparative neuroscience research. The identification of how the anatomo-functional organization of various brain networks evolved through primate evolution is the Grail to understand how each primate species’ brain supports an appropriate behavior repertoire adapted to a specific ecological niche. DYNADAPT’s results will show how networks involved in behavioral adaptation evolved from macaque, chimpanzee, to human. The fourth impact will occur in artificial intelligence and robotic research. One current mainstream international innovation program is to develop autonomous robots assisting humans in their daily life, which are currently almost inexistent. Designing control systems for autonomous robots is extremely difficult because it requires prediction of how the robot will interact with its environment. Recent research has shown the critical potential of neurophysiologically inspired models in such prediction. Data from DYNADAPT will be of major interest for refining these models, and consequently for implementing novel autonomous robotic systems.

Amiez C and Procyk E. Midcingulate somatomotor and autonomic functions. Handbook of Clinical Neurology 166:53-71, 2019.

Amiez C, Sallet, J, Hopkins WD, Meguerditchian A, Hadj-Bouziane F, Ben Hamed S, Wilson CRE, Procyk E, Petrides M. Sulcal organization in the medial frontal cortex provides insights into primate brain evolution. Nature Communications 10(1):3437, 2019.

A hallmark of our survival in the real world is our ability to show behavioral adaptation. Adaptation of behavior can be necessary for a number of reasons, making the study of the process challenging. Two classes of event can signal a need for adaptation: 1) Events caused by one’s own actions and specifically FeedBack –FB– from those actions (e.g. we adapt our strategy after an erroneous choice), and 2) Events not linked to our actions, specifically Action-InDependent Events –AiDE– (e.g. we adapt our strategy after a change of rule). These two types of events frequently occur concurrently and a critical part of adapting appropriately involves resolving the difference between the two. Our task is made even more complex by the fact that the dynamics of evidence accumulation after FB vs AiDE are very different. So, the crucial dilemma is this: after an unwanted outcome, should we adapt as if we made an error and received a negative FB, or should we continue to accumulate evidence as if there has been an AiDE to which we need to know how to adapt. Primates in general and humans in particular are able to resolve this credit assignment problem. Through the primate order, this ability has evolved together with particular brain networks including the frontal cortex. In human, a breakdown of this ability to link unexpected events to their correct cause would seem to be at the source of impairments in a wide range of psychological and neurological disorders. Yet the neural basis of this process and how it evolved through primate evolution are currently unknown. DYNADAPT aims to provide unprecedented characterization of 1) the mode of functioning of brain systems critically involved in this process and 2) their evolution through the primate order to reach its highest level of complexity in human.
DYNADAPT is based on knowledge that I acquired recently in humans and non-human primates and makes the following hypotheses: 1) Relationships between FB, AiDE, and behavioral adaptation are built through learning within a frontal network including the midcingulate cortex (MCC) and the dorsolateral prefrontal cortex (DLPFC) and depend on the integrity of this network. 2) The anatomo-functional organization of this network has evolved from macaque, chimpanzee, to human in such a way that chimpanzees display a very similar organization to humans whereas macaques display all the first-fruits of this organization, strongly suggesting that i) macaque are good models to assess the causal role of this network on behavioral adaptation and on large-scale network functioning, and ii) understanding the organization of this network in chimpanzee is a critical step to bridge the gap between macaque and human and to grasp what makes the human brain unique.
DYNADAPT will demonstrate the validity of these hypotheses in both human and non-human primates thanks to 4 approaches: 1) the various nodes of the network involved will be uncovered in human thanks to functional magnetic resonance imaging (fMRI) and a new adaptive task. 2) the macaque and chimpanzee homologue of these human nodes will be identified thanks to the coupling of RestingState-fMRI (RS-fMRI) and connectivity matching fingerprint analysis. 3) the causal role of these nodes on a) whole brain activity and b) adaptive behavior will be identified in macaque thanks to the coupling of local pharmacological perturbations in these nodes and a) RS-fMRI and b) the assessment of deficits in performance in the adaptive task, respectively.
DYNADAPT will therefore provide critical insights into how the MCC-DLPFC network is organized to produce optimal adaptive behavior through the primate order to reach its highest level of organization in human. As such, it will provide a breakthrough in heterogeneous domains in fundamental, comparative, applied cognitive, and clinical neurosciences.