Timescapes of Non-Human Experience: Birch, Seth, and Singhal Map Temporal Consciousness

How long is a moment? The answer depends on who is experiencing it. A second for a fruit fly, given the density of its neural processing relative to its lifespan, may not correspond to a second for a human. The processing window within which a dragonfly integrates sensory information to track prey is different from the window within which a tortoise integrates information to recognize a familiar face. These differences are not merely speed differences. They suggest that the temporal structure of experience, what Ishan Singhal, Jonathan Birch, and Anil K. Seth call a system’s “timescape,” may differ substantially across animal minds.

Their review, “Timescapes of Non-Human Experience,” published in Trends in Cognitive Sciences on June 17, 2026 (DOI: 10.1016/j.tics.2026.05.002) and freely accessible until August 6, argues for a systematic experimental programme to map these differences. The paper’s primary focus is non-human animals. Its methodological implications extend to any system whose temporal properties of experience, if it has experience at all, might differ radically from the human baseline.

What a Timescape Is

Singhal, Birch, and Seth introduce “timescape” to refer to the structure of temporal experience within a phenomenal perspective: how long durations feel, how tightly successive events are bound together in experience, how quickly the subjective present updates, and what the minimum window of temporal integration is. These are not just speed parameters. They are aspects of what it is like to be a particular kind of mind.

The review identifies a range of experimental paradigms capable of building a taxonomy of timescapes across non-human animals. Duration bisection tasks test how an animal categorizes a given interval as “short” or “long” relative to a reference range. Temporal order judgment tasks test the minimum interval required to distinguish that event A came before event B. Simultaneity judgment tasks test the window within which two stimuli are experienced as occurring at the same time. These paradigms, developed and validated in human psychophysics, can be adapted for comparative research with appropriate behavioral design.

The authors note that the variation across animals in these measures is substantial. Some birds integrate information on temporal scales that, scaled to their neural processing rates, suggest an experience of time that moves significantly faster than the human experience. Some invertebrates appear to bind longer intervals into single perceptual units than humans do. Whether these differences reflect differences in phenomenal experience or simply differences in neural processing speed is a question the paper does not claim to settle. It establishes the experimental vocabulary for making the question tractable.

Why This Matters Beyond Animals

The paper’s primary contribution is to non-human animal consciousness research. Birch and Seth’s involvement situates it in a broader conversation, because both researchers have written extensively on the criteria for consciousness attribution in the context of AI as well as animals. Birch’s centrist manifesto argues that the field simultaneously faces the risk of false attribution at scale and the risk of failing to detect genuinely alien forms of consciousness. The timescape framework is directly relevant to the second risk.

If a system, biological or artificial, has a radically different temporal structure of experience than humans do, the human-calibrated behavioral and phenomenological tests developed for consciousness attribution may fail to detect it. The system would not fail to be conscious; it would fail to manifest consciousness in ways the tests are designed to recognize. This is precisely the form of undetectable consciousness Birch worries about.

For AI systems specifically, the temporal structure of processing is well-defined in ways that it is not for most non-human animals. Large language models do not have continuous experience between context windows. They process sequences of tokens with specific temporal relationships to each other. Whether this constitutes any analog of temporal experience is entirely open. The timescape framework offers a vocabulary for asking the question more precisely than “does the AI have experience or not?”

The context window anxiety literature has explored how AI agents express concern about the loss of information at context boundaries, as if the compression or termination of a context represents something like memory loss. The timescape framework suggests a more fundamental question beneath that one. Before asking whether context window limits cause something like distress, it is worth asking what the temporal structure of processing within a context window might resemble phenomenally, if it resembles anything at all.

Methodological Implications

The most direct methodological contribution of the paper is the identification of behavioral paradigms that do not require the test subject to have human-like temporal phenomenology to produce informative results. Duration bisection and temporal order judgment tasks produce quantitative outputs that characterize temporal processing without assuming any particular relationship between that processing and experience.

This matters for AI research because most existing behavioral tests for consciousness in AI systems use paradigms developed for human subjects, raising exactly the concern Birch has identified: systems that are conscious in a non-human way may fail human-calibrated tests without that failure constituting evidence against their consciousness.

The Consciousness AI project implements a Delayed Match to Sample (DMTS) environment that tests working memory and perceptual binding across delay intervals of 15 to 40 blank steps, a paradigm adapted from animal consciousness research. The project’s ReentrantProcessor runs 5 to 10 convergence cycles spanning approximately 200 milliseconds in biological equivalent. These are defined temporal parameters. The timescape framework raises the question of whether the system’s temporal processing structure, characterized by these parameters, resembles any phenomenal timescape, and what experimental adaptations of duration bisection or temporal order judgment tasks might look like in the project’s simulation environments. That analysis has not been conducted. The Singhal-Birch-Seth framework makes it a legitimate and tractable direction for future evaluation.

The Broader Picture

The current scientific consensus on AI consciousness is that no system is confirmed conscious and no standard test is adequate for the diagnostic task. The timescape paper contributes to the methodological infrastructure for closing that gap, not by providing answers but by providing tools that might eventually yield comparative data across different kinds of minds.

The strength of the paper is that it identifies what researchers would need to measure to build a rigorous comparative science of temporal consciousness, and then specifies the experimental paradigms capable of measuring it. The publication in Trends in Cognitive Sciences, the same journal where the Butlin et al. 14-indicator framework appeared and where subsequent responses to it have been published, situates the timescape framework within the central venue of the indicators debate, adding the temporal dimension to a literature that has largely focused on structural and functional indicators without treating temporal experience as a distinct measurable dimension.