Shedding Our Substantial Baggage: Towards a Process-Ontological Turn in Cognitive Science

Abstract

In the fallout of the replication crisis, several authors have pointed to an underlying "theory crisis" in the cognitive sciences, which are argued to lack a core set of shared principles, assumptions, and methodologies. In this essay, I contend that one major barrier to an integrated field stems from a widespread, often implicit ontological commitment among scientists to substance ontology: the metaphysical view that reality is composed of one or more fundamental, static, independent entities. I argue that this metaphysical assumption contributes to the impression that alternative theoretical decompositions of systems are logically incompatible. I consider cognitive, cultural, and historical reasons for the little-challenged dominance of substance thinking in scientific thought, but suggest that the philosophical foundations are shaky. As such, the field may benefit from engaging more deeply with alternative metaphysical traditions that can be grouped as "process ontologies": the family of views that tend to deny the idea of a fundamental level, take existence to be inherently relational, and emphasize that nature is inherently dynamic. I offer a brief survey of Western philosophical traditions that speak to one or more aspects of process thought, but shine a spotlight in particular on useful concepts from Buddhist and Daoist philosophical traditions. I consider three necessary aspects of all living systems that cannot be accounted for within substance ontology, but can by process: (1) fuzzy boundaries, (2) causal loops, and (3) change. I conclude by suggesting that process ontology may help the field of cognitive science progress by dissolving ill-founded debates regarding the fundamental level of analysis, opening a path for a rigorous meta-theory of cognition.

Cognitive Science Progresses Toward Interactive Frameworks

“…Gibson, 1979;Raja & Anderson, 2019;Spivey, 2020;Turvey, 2018). Similar to how relational quantum mechanics may be changing what physicists look for in superposition and non-locality effects (Candiotto, 2017;Martin-Dussaud, Rovelli, & Zalamea, 2019; see also Mermin, 1998), a process ontology would dramatically change what it is that cognitive scientists are looking for when they hunt down the fundamental elements of cognition (Bickhard, 2003;Falandays, 2021;Rescher, 1996;Seibt, 2009). And thus, as suggested back in Fig.…”

“…Rather than the fundamental operands of cognition being stable representational units inside a brain (Churchland & Sejnowski, 1994), perhaps they are instead information‐bearing processes that emerge in the interaction between an organism and its environment (Bickhard, 2019; Chemero, 2011; Di Paolo, Thompson, & Beer, 2022; J. J. Gibson, 1979; Raja & Anderson, 2019; Spivey, 2020; Turvey, 2018). Similar to how relational quantum mechanics may be changing what physicists look for in superposition and non‐locality effects (Candiotto, 2017; Martin‐Dussaud, Rovelli, & Zalamea, 2019; see also Mermin, 1998), a process ontology would dramatically change what it is that cognitive scientists are looking for when they hunt down the fundamental elements of cognition (Bickhard, 2003; Falandays, 2021; Rescher, 1996; Seibt, 2009). And thus, as suggested back in Fig.…”

On multiscale analyses of neural processing, motor movement, and cognition

“…Stability conditions thus arise at a higher level of analysis that can provide constraints on activity at the lower level of analysis, thereby making a purely bottomup account untenable. In the contextual emergence framework, a linear mechanistic form of reductive analysis will fail to elucidate an emergent property like cognition, however, a form of analysis that allows for multiscale constraints to have mutual influences on one another has promise for decomposing an emergent property into its constituent nonlinear interactive and recurrent processes, if not its constituent parts (Bickhard, 2009;Falandays, 2021;Kallfelz, 2006). Corbin et al (2023) report a clear example of this kind of rich contextual information (which is not always easily labeled) in the motion capture data of sit-to-stand actions, where upcoming intended arm actions are detectable in the postural movements that precede the initiation of those arm actions.…”