Gut-hippocampal memory

The functional argument

Before the mechanism, the reason to expect one. Memory processes are “crucial to foraging and eating behaviors”: an animal must return accurately to a food source and recall episodic features of feeding events — whether a food was nutritive, what time of day and season it was, what the predator situation was. So one should expect the central systems receiving interoceptive energy-status signals to be tightly linked, anatomically and functionally, with the systems that remember the environment.

Quigley et al. (2021) argue they are, at the hippocampus — a structure the wiki otherwise meets only as a memory region in the trauma material.

The evidence, in the order it convinces

The causal rodent result is the strongest thing here. Rats with selective ablation of vagal sensory inputs from the upper gastrointestinal tract are impaired on hippocampus-dependent visuospatial working memory and contextual episodic memory (Suarez et al. 2018). Cut a specific interoceptive afferent, lose a specific memory function — the kind of manipulation the wiki’s cardiac literature cannot perform in humans and rarely performs at all.

The reverse lesion runs the same way. Rats with selective hippocampal lesions cannot use interoceptive energy state (0 vs 24 h food restriction) as a discriminative stimulus for reinforcement — that is, the interoceptive signal stops functioning as information. Human amnesics with non-selective bilateral medial-temporal damage correspondingly fail to adjust hunger ratings after eating and will consume multiple consecutive meals (Rozin et al. 1998; Higgs et al. 2008). The interoceptive signal presumably still arrives; what fails is its use.

Anatomy and physiology.

• Gastric distension and intestinal nutrient infusion increase hippocampal cerebral blood flow in rodents; gastric vagal nerve stimulation does so in humans (Wang et al. 2006).
• VNS induces hippocampal long-term potentiation in freely-moving rats and raises neurotrophic/neurogenic marker expression.
• Receptors for leptin, ghrelin, GLP-1 and insulin are expressed in hippocampus, fluctuate with energy status, and their direct action on hippocampal neurons enhances hippocampus-dependent learning and memory.

Ghrelin is the informative case. It is singled out because it signals energy need — a peripherally derived meal-anticipation signal — not satiation. So memory enhancement is not simply a post-prandial or satiety effect; both ends of the energy cycle modulate the hippocampus. Ghrelin also acts in hippocampus as an interoceptive signal of energy need in otherwise food-sated rats (Suarez et al. 2020), which is as close as this literature comes to showing an interoceptive signal substituting for a bodily state.

The interpretive frame

Suarez et al. (2019), via Quigley et al.: meal-derived vagal signalling primes the hippocampus to encode the mnemonic details associated with eating — where the food source was, the episodic particulars — so as to direct future feeding efficiently.

This is a multisensory-integration claim in the paper’s widened sense: interoceptive energy-status signals from the gut integrated with exteroceptive visuospatial navigational and contextual cues, in one structure, to one functional end.

Why it matters to this wiki

Three reasons, none of them about emotion:

1. It is the first interoceptive function here that is neither affective nor clinical nor about the self. Every other function the wiki records routes through feeling. This one routes through encoding.
2. The evidence grade is better than the wiki’s norm. Selective afferent ablation with a behavioural readout is a stronger design than anything in the cardiac-perception literature, and it exists because the work is in rodents — the trade-off Quigley et al. name as their first theme.
3. It supplies a non-emotional test bed for the general claim that interoceptive signals shape cognition. If interoception is a basic function of the nervous system and not specific to emotion (Barrett’s claim, restated in this same review — see theory-of-constructed-emotion), then it ought to show up in systems with no affective remit. Here it does.

Limits

Almost all rodent. The human contribution is two amnesic-patient observations (behavioural, non-selective lesions, small) and one gastric-stimulation imaging study in obese subjects. No human causal manipulation exists, for the reason given everywhere else in this literature: the afferents cannot be selectively reached. The generalizability caveat Quigley et al. attach to animal models applies here in full, and the review says so — animal work is “instrumental in motivating researchers to develop new experimental methods,” which is a claim about method development rather than about human physiology.