The nervous system is a product of evolution, that is, it was constructed through a long series of modifications, within the strong constraints of heredity, and continuously subjected to intense selection pressures

Neuroscience needs evolution. Paul Cisek and Benjamin Y. Hayden. Philosophical Transactions of the Royal Society B: Biological Sciences, December 27 2021. https://doi.org/10.1098/rstb.2020.0518

Abstract: The nervous system is a product of evolution. That is, it was constructed through a long series of modifications, within the strong constraints of heredity, and continuously subjected to intense selection pressures. As a result, the organization and functions of the brain are shaped by its history. We believe that this fact, underappreciated in contemporary systems neuroscience, offers an invaluable aid for helping us resolve the brain's mysteries. Indeed, we think that the consideration of evolutionary history ought to take its place alongside other intellectual tools used to understand the brain, such as behavioural experiments, studies of anatomical structure and functional characterization based on recordings of neural activity. In this introduction, we argue for the importance of evolution by highlighting specific examples of ways that evolutionary theory can enhance neuroscience. The rest of the theme issue elaborates this point, emphasizing the conservative nature of neural evolution, the important consequences of specific transitions that occurred in our history, and the ways in which considerations of evolution can shed light on issues ranging from specific mechanisms to fundamental principles of brain organization.

4. The importance of major transitions in our evolutionary past

Evolution may occur slowly, but it can have a great effect. Across generations, it can produce large alterations with corresponding adaptations. In addition to papers that emphasize the conservative nature of evolution, our issue includes work that describes several major transitions that took place along the human lineage and made us what we are today. One significant example is the transition from an aquatic to a terrestrial environment. Malcolm MacIver & Barbara Finlay [44] discuss what this meant for sensory systems, especially vision. Due to the properties of light diffraction in water versus air, upon getting out on land our ancestors encountered a visual world that expanded dramatically, by a factor of a million in terms of sensed volume [45]. This offered a vast expansion of opportunities for navigation, as well as decision-making and planning. But there were also new challenges, such as the need for multi-joint limbs and the circuitry to control their movement and posture. All of this produced a great deal of neural expansion and diversification, leading to specific innovations that we find in extant animals, including ourselves.

Other implications of the water to land transition are discussed by Lucia Jacobs, whose paper proposes how air-breathing set the stage for hippocampal evolution in terrestrial tetrapods [46]. Her olfactory navigation hypothesis [47] suggests that olfaction is not just about odour identification, but fundamentally about using odours for spatial navigation. When our ancestors emerged onto land, olfactory sampling became linked with respiration, and Jacobs proposes that this can explain hippocampal theta rhythms, how they could be used to keep track of distance and ultimately for scaffolding mammalian memory.

Another dramatic transition started with the mammalian retreat into nocturnal life and then, about 200 million years later, a return to diurnal life in some primate species. This is described in a paper by Jon Kaas, Hui-Xin Qi and Iwona Stepniewska, which focuses on the corresponding changes to the visual system [48]. In particular, unlike other mammals, primates evolved good vision even when still nocturnal. This was made possible by their large, frontally facing eyes, as well as by a shift in the balance of visual projections to the neocortex, reducing the pathway through the superior colliculus and expanding the more direct retino-geniculo-striate pathway. This was followed by an expansion of the dorsal stream of visual processing into a wide variety of action-specific domains in parietal and premotor regions.

The paper by Paul Cisek summarizes many of these transitions, following along our lineage from chordate filter feeders to mobile aquatic vertebrates, terrestrial tetrapods, nocturnal mammals and diurnal primates [49]. Instead of framing the associated neural innovations as the superposition of new circuits at increasing levels of a hierarchy, with primate cognition at the top, he describes them as the progressive elongation of a general feedback control circuit that gradually subdivided into finer and finer control systems. That is, the highest level of the control hierarchy is the most ancient ‘hypothalamic’ regulation of behavioural state, within which new subdivisions such as abstract planning appeared as adaptations that extended control further into the world and toward more abstract interactions. The resulting architecture, he suggests, retains an ancestral organization into parallel control systems dedicated to guiding particular species-typical actions. Selection between these systems is governed by the basal ganglia, while a selection of specific actions within the chosen system occurs through a competition within each specific cortical map.

Giovanni Pezzulo, Thomas Parr and Karl Friston echo some of these points, emphasizing feedback control as the fundamental organization of the nervous system, but extend it with predictive processing [50]. In particular, they emphasize that predictive processing is by no means a recent evolutionary innovation, but rather a basic principle of vertebrate neural organization that was elaborated from allostatic control to multiple sensorimotor loops that extend in terms of both spatial hierarchy and temporal scales. In this view, cognitive abilities are not added as a new system on top of an old sensorimotor controller, but rather emerge as an extension that specializes part of it toward increasingly abstract and long-term control.

A different but compatible perspective is offered by David Leopold and Bruno Averbeck, who discuss how the vertebrate brain trains itself, a process they refer to as ‘self-tuition’ [51]. They propose that hypothalamic systems modulate telencephalic systems to bias them toward learning the types of information needed for basic functions such as feeding, seeking mates and escaping from threats, as well as orienting and navigating around the world. The complexity of the primate brain, they propose, reflects the complexity of such interactions.

A still more general theoretical treatment of similar issues is offered by Stuart Wilson and Tony Prescott, who define a mathematical framework for how layered control architectures operating at different temporal scales can coordinate to produce complex behaviour [52]. Importantly, while it is widely acknowledged that slower processes can provide the constraints on faster ones, these authors show how the inverse can also be true. The result is a control architecture without a strict hierarchy, but where different levels mutually constrain each other.