Body-Self Modeling, Interoception, and Gender Incongruence: A Predictive-Processing Hypothesis

Persistent gender incongruence involves relationships among experienced gender, bodily phenotype, and representation of the body as self, but these relationships are often compressed into broad constructs such as body image, dysphoria, or interoception. Experimental embodiment research demonstrates that body ownership depends on multisensory integration and can, under important laboratory conditions, be modeled by Bayesian causal inference. Visual, tactile, proprioceptive, interoceptive, and motor information can influence whether a body or body part is experienced as one's own, while sensory uncertainty and prior common-cause probability influence ownership judgments. Transgender-specific research independently indicates that gender congruence can influence own-body identification and virtual embodiment, and that body-self measures and selected neural correlates can change during gender-affirming hormone therapy. Direct evidence that transgender people possess generally altered objective interoceptive accuracy, neural prediction-error signaling, or Bayesian precision weighting, however, is not established. This paper proposes the Gendered Body-Self Inference Model (GBSIM) as a falsifiable computational framework for embodied gender congruence, not as an established theory of the developmental origin of gender identity. Let \(G_t\) denote a gendered self-model, \(Z_t\) a latent inferred body-self state, \(Y_t\) multimodal bodily evidence, \(H_t\) developmental history, and \(M\) the assumed computational model class. The central hypothesis is: \(p(Z_t Y_t,G_t,H_t,M)∝ p(Y_t Z_t,M)p(Z_t G_t,H_t,M)\). A model-defined Gendered Body-State Discrepancy is introduced: \(D_G(t)=∑_i π_i(t)d_i[y_i(t), y_i(t)]\), where \(π_i\) is an effective reliability/precision parameter rather than an empirically established neural precision signal. GBSIM predicts that independently estimated gender-conditioned bodily expectations should improve prediction of body-gender congruence beyond sensory, affective, and explicit cognitive alternatives. The framework introduces nine methodological constraints, competing models, a null-evidence standard, and a preregisterable experimental program capable of rejecting its central propositions. The core boundary is: mechanism of present embodiment ≠ mechanism producing gender identity.