Abstract This paper challenges the “sequence bottleneck” hypothesis, which claims that limitations in sequence memory constrain animal cognition and that human language emerged by overcoming this universal limit. Drawing on comparative neuroscience, behavioral ecology, and language evolution research, we argue that long, linear sequences in human language are not the generative core of linguistic capacity but a niche-specific output format. Across taxa, the hippocampal system — and its analogues — supports the generation of hierarchical, multimodal event structures that are flexibly adapted to each species’ ecological demands. Human language arose when these ancient hierarchical generators were co-opted into a multimodal communicative system shaped by the “linguistic niche,” where pressures for precise, rapid, and cumulative transmission favored the linearization of complex conceptual structures. We highlight three core critiques of the bottleneck view: (i) hippocampal sequence generation is evolutionarily widespread and hierarchically organized, not inherently limited in length; (ii) laboratory sequence tasks often lack ecological validity, underestimating non-human capacities; and (iii) many species achieve sophisticated communication through non-linear, multimodal formats. We propose an integrative neuroecological model in which sequence form is contingent on niche-specific adaptive pressures, reframing cross-species comparisons around the diversity of representational formats rather than deficits in linear output. This approach replaces a deficit-oriented metric with a framework that recognizes both the deep continuity of neural mechanisms and the adaptive divergence of communicative solutions across species.
Edward Ruoyang Shi (2026) studied this question.