Why Are We Hackable? Three Channels of Infiltration in Learning Systems: From Viruses to Algorithms

The article "Neurophysiology of Possession" demonstrated that biological agents (the rabies virus and the Toxoplasma parasite) and memetic agents (extremist ideologies and digital algorithms) redefine host behaviour in favour of their own propagation, doing so through shared neural regions. That article documented the phenomenon of "functional convergence", yet left a more fundamental question unanswered: where does this vulnerability come from, and why do these points of infiltration exist in the brain at all? The present article is built upon that very question and advances three specific claims. First, susceptibility to possession is the inevitable consequence of an adaptive architecture. The same three properties that made social learning and inter-species coexistence possible within the Environment of Evolutionary Adaptedness (EEA), namely open input, adaptive learning, and differential weighting of sources, also constitute the substrate through which possessing agents infiltrate the system. Second, and this is the central claim of the article, all documented cases of possession (biological, chemical, cultural, and algorithmic) pass through at least one of three specific neural channels: 1. Manipulation of prediction error weighting within top-down cortical systems; 2. Fabrication of credible-source signals through prestige bias; 3. Hijacking and commandeering of the mesolimbic reward system. This is a falsifiable claim: finding a single documented case of possession that passes through none of these channels would refute the model. Third, what the entire spectrum of possessing agents shares, from the rabies virus to the digital algorithm (despite their radically different natures and behaviours), is that all of them exploit differing combinations of these same three channels. The distinction between these agents lies not in their point of entry, but in the particular combination of channels they employ. Evidence has been drawn from four layers: biological (rabies, Toxoplasma), chemical (alcohol, cocaine, opioids), memetic and cultural (the domestication of dogs, prestige bias), and convergence with non-living systems (large language models, the foreign-language effect). The evolutionary mismatch framework explains why these channels (which were adaptive within the original environment of evolutionary adaptedness) have become pathological vulnerabilities when confronted with modern artificial stimuli. The article closes with an open question: are "learning" and "susceptibility to possession" inherently inseparable?