Very few papers attempt to overturn over 100 years of thinking about how cognitive sciences should be organised, and even fewer succeed. The article by Paul Cisek in the #time4action special issue of the Psychonomic Society’s journal Attention, Perception, & Psychophysics takes aim at the overarching division of cognitive sciences into the chapters we learn from in our undergraduate textbooks – namely, perception / attention / memory / decision making / action – and argues that this is the wrong way to categorise our study of the mind.
Similar arguments have been made in other recent articles promoting embodied and enactive cognition, but few provide the same breadth of data in support.
The starting point for Cisek’s article is evolution and the behaviour of the first primitive animals that interacted with the world. For example, a simple cephalate (ancestor of all vertebrates) must be able to approach food and to avoid predators. Two distinct neural circuits can be identified for these two behaviours, which have different pathways and different decision processes. This is because, during an Approach action, the creature must select one of the two food sources, while during an Avoid action, it must move away from both. In the neural circuits of this simple creature, there is no box for perception, memory, attention or decision making; rather both neural pathways and decision processes are segregated according to behaviour.
Building on this model of a simple cephalate, Cisek describes how the mammalian cortex can be understood as two interacting processing streams, one for detecting and evaluating stimuli, and the second for specifying actions. The former includes ventral stream pathways and the orbitofrontal cortex, while the latter includes the dorsal stream pathways leading to parietal and premotor cortex. Thus, this model integrates earlier ideas of two visual systems and of affordance competition into an integrated map of how the brain is oriented for controlling behaviour. These cortical systems build on and work with subcortical systems for action control in a way that makes sense of the evolutionary past.
To place this model in a broader context, it is useful to look at another curious example of vertebrate evolution: the sea squirt or tunicate. These creatures have a larva stage in which they look like a tadpole and use their small nervous system to swim to a suitable rock. On finding a rock, they fix in place and digest their own brain and muscles, assuming an entirely passive adult life. The behaviour of this creature is often used as an extreme example of the motor-chauvinist principle that the brain exists in order to allow us to move in a world full of choices. For organisms which don’t move, including plants and the adult sea squirt, a brain is unnecessary.
The model of brain function put forward in Cisek’s article follows the motor-chauvinist idea that the core function of the brain is to control behaviour in relation to the environment. This contrasts with predictive or reward based models, and suggests that the brain is organised according to the need to select actions (based on stimuli in the environment and their values), and the need to specify those actions in detail, with different circuits for different types of action (grasp / orient / vocalise / move). In this model, there is no central ‘decision making’ or ‘attention system’ but rather different behaviours which may come into play in different contexts.
It might seem like quite a leap to go from the neural systems of simple cephalates or sea squirts to a fully-fledged model of the human cortex, and Cisek acknowledges that more work is needed to express all cognitive psychology within this phylogenetic framework.
The paper hints at ways to understand cognitive maps and motivation within this framework. It even recasts the philosophical debate over the nature of representations in these terms, but in both areas, more detailed arguments will probably be needed to reach a settled understanding of how rich cognitive systems can be built on mechanisms that could plausibly evolve.
I would be particularly interested in seeing how social behaviours such as flocking or mating or parenting could be understood in terms of the evolution of action selection mechanisms, and whether such a model could contribute to the development of social neuroscience.
Finally, a new framework for understanding the evolution of brain and behaviour deserves a good name, perhaps one that trips off the tongue more easily than Resynthesizing behaviour through phylogenetic refinement.
Psychonomics article focused on in this post:
Cisek, P. (2019). Resynthesizing behavior through phylogenetic refinement. Attention, Perception, & Psychophysics. DOI: 10.3758/s13414-019-01760-1.