Phd in PHD PROGRAM IN BEHAVIORAL NEUROSCIENCE

Thesis title: Decision making driven by mental schemas: study of neuronal correlates of a transitive inference task in dorsal prefrontal and premotor cortex of primates

In several circumstances, decisions rely on a mental model of previously learned and connected information or events. If we learned that A > B in a given time, and in another moment, we learned that B > C, we can deduce that A > C even if were never observed together. This decision is possible because we implicitly organize related information on a metal representation where items are rank-ordered as A>B>C. This way to represent knowledge supports decision-making in several contexts, including inferential reasoning. Besides humans, many other species have been shown to rely on this ability to decide, including those requiring the solutions of Transitive Inference (TI) problems in different contexts. Many experimental models have been designed to study the cognitive and neuronal basis of TI and the mental organization of related information in humans and other animals. In this thesis, we explored the neuronal basis of decision-making during a non-verbal form of TI task in non-human primates. Previous neurophysiological studies have reported that neuronal activity in the Prefrontal cortex is modulated during inferential reasoning while an abstract mental schema of ranked items is accessed. In addition to PFC, recent studies in primates have reported a greater involvement of the premotor cortex in cognitively demanding tasks, such as TI. The neurons in the premotor cortex have been shown to encode abstract rules for selecting proper action. With this pretext, we compared the neuronal modulation of dorsal prefrontal (DLPFC) and premotor cortex (PMd) in monkeys while they performed a TI task during which they were required to choose the higher ranking item from a pair of abstract images, selected from a rank-ordered series (A>B>C>D>E>F). The monkeys had previously learned the adjacent relationships between the series items during the learning phase (e.g., A>B; B>C), while they were tested with ordinal pairings not presented to them during the learning (e.g., A vs C). During the decisions on these novel pairings, we observed that the choice accuracy increased and the reaction time decreased as the rank difference between the paired items increased. These results are evidence of the construction of an abstract mental schema by the monkey to solve the task. On a neuronal level, both areas exhibited a higher neuronal activity when the target item appeared at a specific spatial location as compared to the non-target item, and this difference was more prominent at lower degrees of difficulty. Furthermore, a comparison of the time evolution of this activity revealed that the target item position was encoded earlier in by DLPFC than PMd neurons. These results, along with previous neurophysiological studies, show that the neuronal activity in DLPFC is modulated when an abstract mental representation of ranked items is accessed during the test phase of a TI task. However, the question of how the neuronal encoding of the individual ranked items leading to this representation is shaped through learning is relatively unexplored. We addressed this question in the latter part of the thesis by investigating the single-cell activity from DLPFC during the learning phase and comparing this encoding of ranked items with the corresponding neuronal modulations in the test phase. As per our results, the DLPFC neurons indicate that the learning of the proximate relations between the pairs of items reshaped the neuronal encoding of the items’ rank and supported the representation of the item during the test phase. The results from both the studies presented in this thesis provide evidence of the involvement of DLPFC and PMd in decision making driven by mental schema and a role of DLPFC in encoding this schema to support flexible manipulation of serial information for taking decisions.