Veissière et al. have uncovered an interesting set of high-level regularities, which appear to show up wherever humans attempt to calibrate their behavior against one another. They have also shown that the FEP provides a unified mathematical framework that is useful for describing these regularities. Highly general and unified explanatory models such as TTOM can be extremely useful. For example, where a discipline lacks a common theoretic language for describing competing perspectives, such models can be deployed to dissolve disputes by bringing rival positions under a single theoretical framework. Veissière et al. assume that such a strategy will prove fruitful in resolving persistent disagreements within the cognitive science of cultural acquisition and social cognition, as TTOM seems to provide a unified framework for characterizing insights from a number of otherwise incommensurable theories. We applaud their attempt to provide a more unified account of social cognition and the acquisition of culture; but we contend that bringing these phenomena under a single mathematical framework is unlikely to resolve the relevant disputes.
Providing a simple, overarching characterization of complex and inherently variable biological systems is challenging. Abstract mathematical models of biological phenomena, such as TTOM, attempt to overcome this challenge by prioritizing explanatory generality over competing ideals such as explanatory precision and biological realism (Levins Reference Levins1966). If successful, this strategy can offer unifying explanations of seemingly disparate biological phenomena, such as the action of different but analogous biological systems, or of heterogeneous parts of the same system. Yet, unification comes at the cost of explanatory realism and precision. In order to draw parallels between non-identical systems, general models must make idealizing assumptions about patterns of biological variation as well as the causal specificities of the particular systems being described. This allows such models to capture the general properties of a system, by focusing on broadscale similarities. But as a result, they fail to be entirely precise and accurate when it comes to the particularities of the system (Woodward Reference Woodward2005). The mathematical framework provided by the FEP does give TTOM a high level of generality; but we worry that this involves stripping away fine-grained causal details and evolutionary histories without much obvious explanatory pay-off.
This is not to deny that unification can offer new perspectives, but we doubt that there is more to say about social cognition and cultural acquisition at the highly abstract level afforded by the FEP than is already being said at a less general, but causally richer, level of description. This concern might be mitigated if TTOM succeeded in providing a common framework for usefully describing and comparing competing theories in cognitive science, but we worry that any theoretical unification achieved via TTOM will be more perspectival than substantive, as the unification it provides is generated by looking at the issues from a level of abstraction that makes the details disappear. Long-standing debates in cognitive archeology illustrate these problems nicely.
The story one tells about the evolution of hominin cognition is highly dependent on the position one adopts on social cognition. Debates between dynamicists/externalists (Malafouris Reference Malafouris, Wynn and Coolidge2016; Noble & Davidson Reference Noble and Davidson1996; Overmann Reference Overmann, Wynn and Coolidge2016; Tomlinson Reference Tomlinson2015) and representationalists/internalists (Cole Reference Cole, Wynn and Coolidge2016; Coolidge & Wynn Reference Coolidge and Wynn2018; Mithen, Reference Mithen1996) in cognitive archeology mirror broader debates in cognitive science. For instance, Noble and Davidson (Reference Noble and Davidson1996) employ an externalist and Gibsonian approach to the analysis of stone-tools and the evolution of social cognition, whereas Mithen (Reference Mithen1996) employs an internalist and modular approach. If TTOM provides a tool for resolving debates in cognitive science, it should also offer the resources for arbitrating between these different views, and for finding a clear route to a resolution. Unfortunately, even if TTOM can express these rival accounts in the general, abstract, mathematical language, this redescription seems to add little to our existing, much richer causal understanding of the systems in question.
Debates about hominin cognitive evolution largely concern the kinds of cognitive traits that are required to produce lithic technologies. And resolving such debates requires generating mutually exclusive and testable empirical predictions to compare against the Paleolithic record and findings in contemporary cognitive science; any common vocabulary for comparing theories must be causally rich enough to engage with such evidence. Unfortunately, TTOM is so abstract and multiply realizable that the evolutionary histories and fine-grained causal information that instantiate the competing views about hominin cognitive evolution are largely omitted. Given this causal frugality, TTOM seems incapable of generating the testable predictions cognitive archeologists require to resolve these debates, and hence the overall payoff for deploying it is unclear.
We suspect that the state of affairs in cognitive archeology is a reflection of broader debates in the study of human social cognition and cultural acquisition. Recent experiments have revealed significant intra- and inter-personal variation in mentalizing capacities (e.g., Warnell & Redcay Reference Warnell and Redcay2019); this may reflect heterogeneity in the underlying biological systems (Schaafsma et al. Reference Schaafsma, Pfaff, Spunt and Adolphs2015), or it may suggest the development of different kinds of sense-making strategies (De Jaegher Reference De Jaegher2013). An approach that focused on patterns of qualitative variation might yield empirically tractable predictions in this domain; and given a plausible set of bridging principles, the resulting data may be useful for adjudicating the relevant disputes. By contrast, the unified theoretical framework advanced by Veissière et al. can only reveal the points where these different kinds of approaches are likely to converge. As we see it, TTOM mistakenly equates formal unification with explanatory power. Explanation in science is, alas, far more complex; and generality comes at the cost of valuable explanatory realism and precision. In light of this worry, we contend that the explanatory value which TTOM appears to have is likely to reflect its ability to systematize existing data, rather than its ability to produce novel hypotheses, or novel ways of negotiating intractable disputes.
Veissière et al. have uncovered an interesting set of high-level regularities, which appear to show up wherever humans attempt to calibrate their behavior against one another. They have also shown that the FEP provides a unified mathematical framework that is useful for describing these regularities. Highly general and unified explanatory models such as TTOM can be extremely useful. For example, where a discipline lacks a common theoretic language for describing competing perspectives, such models can be deployed to dissolve disputes by bringing rival positions under a single theoretical framework. Veissière et al. assume that such a strategy will prove fruitful in resolving persistent disagreements within the cognitive science of cultural acquisition and social cognition, as TTOM seems to provide a unified framework for characterizing insights from a number of otherwise incommensurable theories. We applaud their attempt to provide a more unified account of social cognition and the acquisition of culture; but we contend that bringing these phenomena under a single mathematical framework is unlikely to resolve the relevant disputes.
Providing a simple, overarching characterization of complex and inherently variable biological systems is challenging. Abstract mathematical models of biological phenomena, such as TTOM, attempt to overcome this challenge by prioritizing explanatory generality over competing ideals such as explanatory precision and biological realism (Levins Reference Levins1966). If successful, this strategy can offer unifying explanations of seemingly disparate biological phenomena, such as the action of different but analogous biological systems, or of heterogeneous parts of the same system. Yet, unification comes at the cost of explanatory realism and precision. In order to draw parallels between non-identical systems, general models must make idealizing assumptions about patterns of biological variation as well as the causal specificities of the particular systems being described. This allows such models to capture the general properties of a system, by focusing on broadscale similarities. But as a result, they fail to be entirely precise and accurate when it comes to the particularities of the system (Woodward Reference Woodward2005). The mathematical framework provided by the FEP does give TTOM a high level of generality; but we worry that this involves stripping away fine-grained causal details and evolutionary histories without much obvious explanatory pay-off.
This is not to deny that unification can offer new perspectives, but we doubt that there is more to say about social cognition and cultural acquisition at the highly abstract level afforded by the FEP than is already being said at a less general, but causally richer, level of description. This concern might be mitigated if TTOM succeeded in providing a common framework for usefully describing and comparing competing theories in cognitive science, but we worry that any theoretical unification achieved via TTOM will be more perspectival than substantive, as the unification it provides is generated by looking at the issues from a level of abstraction that makes the details disappear. Long-standing debates in cognitive archeology illustrate these problems nicely.
The story one tells about the evolution of hominin cognition is highly dependent on the position one adopts on social cognition. Debates between dynamicists/externalists (Malafouris Reference Malafouris, Wynn and Coolidge2016; Noble & Davidson Reference Noble and Davidson1996; Overmann Reference Overmann, Wynn and Coolidge2016; Tomlinson Reference Tomlinson2015) and representationalists/internalists (Cole Reference Cole, Wynn and Coolidge2016; Coolidge & Wynn Reference Coolidge and Wynn2018; Mithen, Reference Mithen1996) in cognitive archeology mirror broader debates in cognitive science. For instance, Noble and Davidson (Reference Noble and Davidson1996) employ an externalist and Gibsonian approach to the analysis of stone-tools and the evolution of social cognition, whereas Mithen (Reference Mithen1996) employs an internalist and modular approach. If TTOM provides a tool for resolving debates in cognitive science, it should also offer the resources for arbitrating between these different views, and for finding a clear route to a resolution. Unfortunately, even if TTOM can express these rival accounts in the general, abstract, mathematical language, this redescription seems to add little to our existing, much richer causal understanding of the systems in question.
Debates about hominin cognitive evolution largely concern the kinds of cognitive traits that are required to produce lithic technologies. And resolving such debates requires generating mutually exclusive and testable empirical predictions to compare against the Paleolithic record and findings in contemporary cognitive science; any common vocabulary for comparing theories must be causally rich enough to engage with such evidence. Unfortunately, TTOM is so abstract and multiply realizable that the evolutionary histories and fine-grained causal information that instantiate the competing views about hominin cognitive evolution are largely omitted. Given this causal frugality, TTOM seems incapable of generating the testable predictions cognitive archeologists require to resolve these debates, and hence the overall payoff for deploying it is unclear.
We suspect that the state of affairs in cognitive archeology is a reflection of broader debates in the study of human social cognition and cultural acquisition. Recent experiments have revealed significant intra- and inter-personal variation in mentalizing capacities (e.g., Warnell & Redcay Reference Warnell and Redcay2019); this may reflect heterogeneity in the underlying biological systems (Schaafsma et al. Reference Schaafsma, Pfaff, Spunt and Adolphs2015), or it may suggest the development of different kinds of sense-making strategies (De Jaegher Reference De Jaegher2013). An approach that focused on patterns of qualitative variation might yield empirically tractable predictions in this domain; and given a plausible set of bridging principles, the resulting data may be useful for adjudicating the relevant disputes. By contrast, the unified theoretical framework advanced by Veissière et al. can only reveal the points where these different kinds of approaches are likely to converge. As we see it, TTOM mistakenly equates formal unification with explanatory power. Explanation in science is, alas, far more complex; and generality comes at the cost of valuable explanatory realism and precision. In light of this worry, we contend that the explanatory value which TTOM appears to have is likely to reflect its ability to systematize existing data, rather than its ability to produce novel hypotheses, or novel ways of negotiating intractable disputes.