Logical neurons for logical reasoning - Stefano Ferraina

Only recently after the advent of imaging techniques has neuroscience begun to examine the neural fundamentals of logical reasoning—the essence of the human mind. Unfortunately, imaging does not have the power to effect an in-depth understanding of how the brain organizes and manipulates information to infer new knowledge. No neurophysiological data on this topic exist in humans or animals. The common observation that monkeys use inference to learn social dominance strongly supports the hypothesis that they adopt ‘human-like’ reasoning and implicates them as animal model candidates.

Based on neuroimaging and lesion studies, transitive inference (TI), one of the most studied forms of logical reasoning, is likely mediated by the interaction of several areas in the temporal, parietal, and frontal lobes.

We started to explore the issue by testing humans and animals in similar paradigms. Data we obtained show that humans and animals have similar behavioral characteristics and that animals reach optimal performance on various forms of the TI task in a 1-day session. 

To examine the cortico-cortical mechanisms that are used to actively build a distributed representation of logical conclusions in TI, we focused in the parieto-frontal network.  Animals were trained to learn, in a single session, the relationship between six adjacent items (premises) of an arbitrarily rank-ordered sequence (i.e. A>B; B>C; C>D; D>E; E>F; learning phase) and then to deduct the relationship between novel (non-adjacent, e.g., B>D or C>E) pairs never experienced during the learning phase and presented intermingled to learned pairs (test phase). 

Results support the behavioral evidence of a mental line use to organize premises, after learning.