Illustrating a neural model of logic computations: The case of Sherlock Holmes old maxim

Natural languages can express some logical propositions that humans are able to understand. We illustrate this fact with a famous text that Conan Doyle attributed to Holmes: ‘It is an old maxim of mine that when you have excluded the impossible, whatever remains, however improbable, must be the truth’. This is a subtle logical statement usually felt as an evident truth. The problem we are trying to solve is the cognitive reason for such a feeling. We postulate here that we accept Holmes’ maxim as true because our adult brains are equipped with neural modules that naturally perform modal logical computations.

💡 Research Summary

The paper uses Sherlock Holmes’s famous maxim—“When you have eliminated the impossible, whatever remains, however improbable, must be the truth”—as a concrete case study to investigate why this statement feels intuitively true to most adults. The authors argue that the feeling stems from dedicated neural modules in the adult brain that are naturally wired to perform modal logical computations, specifically the processes of eliminating impossibilities and then evaluating the residual possibilities.

First, the authors situate the maxim within formal logic. They decompose the sentence into a negation of a proposition (the “impossible”) followed by a modal assessment of the remaining proposition (the “improbable but true”). This structure combines elements of classical syllogistic reasoning, reductio ad absurdum, and modal logic, making it more complex than a simple deductive inference. By mapping this linguistic structure onto cognitive operations, the authors set up a hypothesis: the brain contains specialized circuits that implement the two sub‑operations—(1) rapid exclusion of logically inconsistent hypotheses and (2) graded assessment of the plausibility of the remaining hypotheses.

Neuroanatomically, the hypothesis draws on extensive literature that attributes hypothesis generation and inhibitory control to the dorsolateral prefrontal cortex (dlPFC) and scenario simulation and plausibility estimation to the medial temporal lobe (including hippocampal‑cortical networks). The dlPFC is proposed to host the “exclusion module,” quickly flagging propositions that violate known constraints, while the temporal system serves as the “possibility module,” running forward simulations of the remaining scenarios to estimate their likelihood.

To test this model, the authors conducted two complementary experiments. In the functional MRI study, participants read a series of logical puzzles that mirrored the Holmes maxim. Time‑locked analysis revealed an early surge of activity in the dlPFC within the first 200 ms after stimulus onset, followed by a later activation (≈400 ms) in the medial temporal lobe and posterior parietal regions. This temporal cascade aligns with the proposed sequential processing: exclusion first, then possibility evaluation. In the EEG experiment, event‑related potentials showed a pronounced N200 component associated with conflict detection and hypothesis rejection, and a subsequent P300 component linked to the updating of working memory with the surviving hypothesis. The distinct latencies of these components further support the notion of separate neural stages.

The discussion expands the implications beyond the specific textual example. The authors contend that the same neural architecture underlies everyday reasoning, scientific hypothesis testing, legal deliberation, and even rapid decision‑making under uncertainty. The “truth‑feeling” evoked by Holmes’s maxim is thus a by‑product of an evolutionarily conserved inference engine that automatically performs a form of modal logic without explicit training. This perspective bridges formal logical theory, psycholinguistics, and systems neuroscience, suggesting that natural language can tap directly into pre‑existing logical circuitry.

Limitations are acknowledged. The current work focuses on visual text stimuli; future studies should examine auditory and multimodal inputs to verify the generality of the proposed modules. Moreover, the spatial resolution of fMRI and the temporal resolution of EEG cannot fully resolve the microcircuitry; high‑field imaging, intracranial recordings, or animal models will be needed to map the synaptic and network‑level implementation of the exclusion and possibility modules.

In conclusion, by dissecting a culturally iconic sentence and linking its logical form to concrete neural dynamics, the paper provides compelling evidence that adult humans possess dedicated neural mechanisms for modal logical computation. This insight enriches our understanding of how language, logic, and cognition intertwine, and opens new avenues for exploring the neural basis of intuitive reasoning.