When the Hexagons Fail: The Entorhinal–Hippocampal Origin of Autism’s Social Disorder

1. Introduction

Autism is defined behaviorally, but its social features form a recognizable and recurring cluster. Reduced declarative pointing, palm-reversed waving, reduced joint attention, and later difficulties in social positioning are usually treated as distinct phenomena, each with its own literature and proposed mechanism. I propose that they are different developmental expressions of the same underlying coordinate-system disorder.

Their co-occurrence is informative. When multiple features reliably appear together, a single shared mechanism is a more parsimonious explanation than several independent ones. I therefore ask which single mechanism, if disrupted early, would produce this specific set of features. I propose the answer concerns how the brain assigns coordinates to space and to other people.

2. The Grid Code and Non-Spatial Representation

The medial entorhinal cortex contains grid cells that fire in a periodic hexagonal pattern during movement through space (Hafting et al. 2005). With hippocampal place cells (O’Keefe and Dostrovsky 1971), they form the brain’s spatial representation system (O’Keefe and Nadel 1978). This system is present in humans, demonstrated indirectly by the six-fold fMRI signal (Doeller et al. 2010) and directly by intracranial recordings (Jacobs et al. 2013).

Critically, this code extends beyond physical space. The same grid-like signal represents abstract conceptual dimensions (Constantinescu et al. 2016) and social ones. When people learn relationships defined by power and affiliation, the hippocampus encodes others as positions along these dimensions (Tavares et al. 2015), and social hierarchy inference recruits the same grid-like code (Park et al. 2021). The spatial coding system is thus reused to represent people (Behrens et al. 2018).

This establishes the premise: the same grid-like computation is recruited to represent relationships between people, not only locations in space.

3. The TPJ as a Downstream Self–Other Processor

Spatial cognition requires transformations between the allocentric maps encoded by the entorhinal–hippocampal system and the egocentric representations used for action. Social cognition requires an analogous transformation—between self-centered and other-centered reference frames. I propose that the TPJ operates on these transformed coordinate representations, and therefore depends on the integrity of the upstream code that generates them.

The temporoparietal junction (TPJ) supports self–other distinction, perspective-taking, and representation of one’s own bodily location (Blanke 2012; Decety and Lamm 2007). These functions depend on spatial and self-referential coordinates.

The TPJ is repeatedly implicated in autism. In one fMRI study, the right TPJ in neurotypical adults showed enhanced activity when tracking another person’s false belief compared to a true belief; in autistic adults, this enhancement was attenuated (Nijhof et al. 2018). Notably, task performance did not differ between groups—the difference appeared at the neural level rather than in overt behavior. This dissociation is consistent with a processing difference that does not necessarily surface as task failure.

Importantly, entorhinal grid-like representations have been shown to relate to TPJ decision computations (Park et al. 2021), providing functional evidence that entorhinal grid-like coding informs TPJ computation.

The conventional interpretation treats the TPJ finding as a primary deficit of the social brain itself. I propose a different causal order: the TPJ abnormality may be secondary—arising because the region operates on distorted input from upstream.

4. The Hypothesis

I state the hypothesis directly:

By “distorted coordinates,” I mean representations that are unstable, imprecise, or misaligned across self-centered, other-centered, and shared reference frames.

The causal direction distinguishes this account. The conventional view holds that the TPJ itself is impaired. On this account, the TPJ abnormality is secondary rather than primary: the region operates on distorted coordinate input from upstream. The implication is that the relevant target is the upstream coordinate distortion rather than the downstream region.

5. Reinterpreting the Symptoms

Each feature can be described as a coordinate-transformation failure.

Palm-reversed waving. Producing a conventional wave requires transforming one’s own hand configuration into an observer-centered reference frame. A distorted coordinate representation leaves the gesture in the child’s own reference frame, palm inward.

Absence of declarative pointing. Pointing to share requires aligning a spatial vector with another person’s reference frame. This is a coordinate-registration operation that fails without a stable shared representation.

Reduced joint attention. Joint attention requires two individuals to converge on the same target within a shared coordinate space. Misaligned representations cannot be matched.

Difficulty placing others relationally. The social representation itself, encoded by the grid code, fails to assign others stable positions along its dimensions.

These four features reduce to a single underlying operation.

6. The Energy Link

The entorhinal–hippocampal system is among the brain’s most metabolically demanding circuits. Maintaining a spatial map through continuous path integration requires persistent neural firing, recurrent network activity, and oscillatory synchronization—all energetically expensive. The entorhinal cortex is also among the earliest regions affected in Alzheimer’s disease, a disorder marked by mitochondrial, metabolic, and network dysfunction. This pattern is consistent with exceptional vulnerability of the coordinate system to sustained energetic compromise.

I have previously proposed that multiple cytokines converge on mitochondrial dysfunction to produce a cerebral energy deficit underlying core features of autism (Kang 2025). A high-demand coordinate system is a plausible selective target of such a deficit. Energy scarcity may therefore be one mechanism by which the grid code becomes distorted.

7. Falsifiable Predictions

1.

Grid-like fMRI signals (six-fold periodicity) will be weaker or less regular in autistic individuals during spatial and conceptual navigation tasks.

2.

The degree of grid-code disruption will correlate with the severity of social-coordinate features, but not with non-social features.

3.

Social-space mapping tasks (power and affiliation dimensions) will show degraded coordinate structure in autism.

4.

Grid-code integrity will predict social-coordinate features—joint attention, perspective transformation, and declarative pointing—more strongly than non-social features such as repetitive behavior, restricted interests, or sensory hypersensitivity.

5.

The association between autism-related social features and TPJ social-inference responses will be mediated by grid-code integrity.

6.

Disrupting entorhinal coordinate coding should alter TPJ social-inference responses even when the TPJ itself is structurally intact.

8. Conclusions

Autism’s apparently separate social features are different expressions of a single coordinate-system disorder. The primary disturbance lies upstream of the TPJ, in the entorhinal–hippocampal code that supplies the relational structure on which self–other computation depends. If this code is the source, identifying what disrupts it is a key step toward the upstream etiology of autism’s social disorder.