This book is a thought experiment. The thought tested is whether the acquisition of the ability to engage in linguistic conversation with others—which the author dubs ‘converse’—can account for the origins of the distinguishing capacities of the human mind. It is arguably one of the most comprehensive accounts of a ‘language-first’ theory of thinking and morality that recent philosophy has produced. At the same time, its hypotheses are testable, so that it could become a wellspring for scientific experiments in developmental and comparative psychology for decades to come.

The distinguishing features of the human mind that are subjected to the thought experiment are six, and it is around these six that the book is organized. Three of these are what the author calls ‘processing’ capacities (we could also call them epistemic capacities): judgement, reasoning, and percipience (chapters 2–4). Each of these, as characterized, involves a degree of meta-cognitive reflection: in judgement, we reflect on whether a claim is supported by evidence; in reasoning, we reflect on whether our beliefs are supported by good reasons; and in percipience, we reflect on whether the deliverances of perception should be accepted as reality. The other three are ‘relational’ capacities (we could also call them moral capacities): endorsing norms and values, holding one another to those values, and engaging with one another as persons (chapters 5–7).

How, then, might ‘converse’—to stick with Pettit’s terminology—play an explanatory role in their origin? There is a common argumentative form to the accounts of why converse leads to the emergence of each capacity. Converse is not (for the most part) considered as a direct cause, or the mechanism, that explains the emergence of the inner capacity. Rather, the practice of converse is argued in each case to create pressure for the inner capacity to emerge. This is the argumentative strategy of a ‘genealogical’ account of the emergence of the mind, which the book is explicit about adopting. The task of a philosophical genealogy is not necessarily to explain the mechanisms that gives rise to each step in the lineage, but to make sense of why each step in the lineage might have preceded the next. The place of converse in these arguments can be compared to the place of environmental changes in accounts of why traits emerge via natural selection. In such accounts, environmental changes are not thought of as the mechanism that causes the new traits to appear in the first place; rather, environmental changes are said to explain why the traits come to dominate in the species, because of the pressure they put on the species to have those traits if it is to survive.

Consider the argument for why converse would result in the emergence of the capacity for judgement. Suppose that agents with the linguistic ability to express first-order statements begin to exchange those statements with one another—to engage in conversation. Since different speakers will have different beliefs, sooner or later speakers will express conflicting beliefs. One speaker may say that the deer are in the valley, but another speaker may assert, contrarily, that the deer are on the mountain. If the two need to collaborate to catch the deer, how are they to do this if they do not resolve the contradiction? The disagreement creates pressure for each speaker to produce evidence to back up their assertion, which they can only do if they adopt ‘evidence-seeking’ into their repertoire of internal cognitive capacities. If they cannot identify and then verbally articulate the evidence that could back up their claims, they will be unable to settle the conflict one way or another, and unable to collaborate to catch the deer. Converse in this way puts pressure on the minds involved to develop the ability to identify the evidence that would be needed to justify their claims. And this kind of sensitivity to evidence in the context of assertions is what Pettit characterizes as the capacity for judgement. Now the agents make assertions only when they are aware that they have sufficient evidence to satisfy the inevitable interrogation they will be subjected to. And this kind of selectivity in deciding what to believe is what Pettit characterizes as the capacity for judgement—not mere belief-formation, but the formation and expression of beliefs only when they are sufficiently well supported.

This general argumentative strategy makes the core arguments of the book resilient to some standard objections to language-first accounts, or accounts that take some aspects of cognition to depend on language. Some have argued that acquiring language directly supports the representational machinery required to think certain thoughts. De Villiers and de Villiers (2014) has argue that it is due to acquiring ‘sentential complement clause’ syntax (a sentence like ‘Sally thinks that the ball is in the box’) that young children acquire the ability to represent another agent as having an attitude to a proposition, or to meta-represent the thoughts of others. On the de Villiers’ account, the internally represented sentence is the vehicle of the meta-representation. Clark (1998) and Bermúdez (2003) have argued, similarly, that a sentence in a language can serve as the mode of presentation of a thought—thereby making meta-cognition of our own thoughts possible. Again, the representation of the sentence is the vehicle of the meta-cognition. Such arguments are vulnerable to a circularity objection: how can a mind represent something by acquiring the use of a word, or string of words, when acquiring the words seems to depend on the mind’s ability to represent the reference of the expression in the first place? The capacity to represent the reference of the expression would need to already exist if the expression were ever to be learned. This is the central argument in Fodor’s (1975) The Language of Thought, which cast itself as a rebuttal of language-first accounts of the emergence of the mind (specifically Wittgenstein 1953). Crucially, Pettit’s argumentative strategy is not vulnerable to this kind of concern. He is not (for the most part) claiming that language functions as the vehicles of representation; rather, he is claiming that the emergence of linguistic interactions created pressure for the inner capacities to emerge, thereby explaining why one thing (such as overt verbal disagreement) might come before another (such as internal sensitivity to evidence).

A growing body of empirical work supports the genealogy of the human mind that Pettit proposes. Consider again the capacity of judgement—the disposition to be sensitive to evidence and to seek it out ‘intentionally with a view to updating or confirming our beliefs’ (p. xiv). We designed an experiment to test for this capacity in young children, either in a context of peer disagreement or in an individual, ‘asocial’ context (O’Madagain et al. 2022). First let’s consider the individual context. Three- and five-year-olds were presented with two pieces of perceptual evidence in succession about the location of a reward (a marble, which could be used to play a game). In some trials, the second piece of evidence contradicted the first. The critical question was whether children would recognize that this contradiction undermined the reliability of any conclusion they could draw and would, therefore, seek additional evidence before making a final decision. Three-year-olds systematically failed to do this. Rather than checking for more information, they based their decision on the most recent evidence, apparently disregarding the earlier, conflicting evidence. At this age, children seem limited in their ability to monitor and integrate contradictory perceptual evidence within their own experience. Now let’s consider the social version of the task. This time one child was given a first piece of evidence, while a partner received a conflicting piece of evidence. When the first child observed the second make a choice that contradicted their own, they were significantly more likely to look for more evidence than they were in the first, individual experiment. This supports Pettit’s proposal that disagreement plays a crucial role in the emergence of judgement: social conflict places pressure on individuals to reflect on and reassess their evidence. Absent such social pressure, the youngest children appeared strikingly insensitive to contradictions in their own perceptual experience (O’Madagain et al. 2022; see also Helming et al. 2024).

However, these results come with a big twist. When great apes participated in the same paradigm, we saw exactly the opposite pattern. Faced with two successive pieces of contradictory perceptual evidence, apes reliably looked for more information before making a final decision—significantly more, indeed, than three-year-old children. But when apes received a single piece of evidence while a partner selected the opposite location, they were significantly less likely to recheck than when they had seen conflicting perceptual evidence—largely ignoring the partner’s ‘disagreement’. For human children, social disagreement is a stronger trigger for evidence-checking than conflicting perceptual evidence—just as Pettit’s genealogy might predict. For great apes, the reverse is true (for further evidence of great apes’ remarkable sensitivity to evidence, including abandoning beliefs when they discover their evidence was faulty, see Schleihauf et al. 2025).

These results seem to both support and undermine Pettit’s story at once. They support the idea that young humans acquire the capacity for judgement due to the pressure exerted by peer disagreement. What they undermine is the suggestion made in several places that converse is necessary, or that humans, being the only species that engages in converse, is the only species with these capacities. In some other species, it seems, it is not social pressure that prompts judgement. Rather, we see evidence for the capacity primarily in contexts of individual decision-making. Indeed, we can think of lots of sources of pressure for such a capacity to emerge, apart from social disagreement. The ability to keep track of and evaluate evidence while making decisions is clearly of benefit to individual survival. Consider a creature that hesitates when it is presented with apparently contradictory information about the location of a prey it is hunting, to double-check before making a final decision. She will be more likely to catch the prey than one who is swayed to choose now one option, now another, without any regard for the incoherence in their choices. In other words, it is not too difficult to think of sources of pressure apart from peer disagreement that might bring about the emergence of a capacity for judgement.

The upshot of these considerations is that there may be multiple routes by which agents may acquire aspects of the life of the mind that humans have (O’Madagain 2026). One route is social. If language emerges in a species that is not sensitive to tracking the evidence that supports its claims, the disagreement between peers this will generate can create pressure that brings about that capacity. But another route is non-social. If an agent is often presented with conflicting or confusing evidence in its search for food, then this too will create pressure for an ability to keep track of and evaluate evidence before forming a final decision.

It is worth bearing in mind, then, that we can identify at least precursors of some of the capacities Pettit discusses in other species, and these appear to have developed independently of social practices. How deep such wordless capacities run remains an open question. Even so, the great achievement of When Minds Converse lies in making sense of the distinctively social route by which these capacities seem to develop in humans, which is often so perplexing.

Cathal O’Madagain
School of Collective Intelligence, Rabat, Morocco
and
Ecole Normale Superieure, Paris, France
cathal.omadagain@um6p.ma

References

Bermúdez, J. L. (2003). Thinking Without Words, Oxford University Press.

Clark, A. (1998). ‘Magic Words: How Language Augments Human Computation’, in P. Carruthers and J. Boucher (eds), Language and Thought: Interdisciplinary Themes, Cambridge University Press, pp. 21–39.

De Villiers, J. G. and de Villiers, P. A. (2014). ‘The Role of Language in Theory of Mind Development’, Topics in Language Disorders, 34, pp. 313–28.

Fodor, J. (1975). The Language of Thought,  Harvard University Press.

Helming, K., O’Madagain, C. and Tomasello, M. (2024). ‘Three-and 5-Year-Old Children Know Their Current Belief Might Be Wrong’, Journal of Experimental Child Psychology, 246, available at .

O’Madagain, C. (2026). ‘Ways of Reasoning in Humans and Other Animals’, in P. Stovall and L. Koren (eds), Why and How We Give and Ask for Reasons: Perspectives from Philosophy and the Sciences, Oxford University Press, pp. 269–90.

O’Madagain, C., Helming, K. A., Schmidt, M. F. H., Shupe, E., Call, J. and Tomasello, M. (2022). ‘Great Apes and Human Children Rationally Monitor Their Decisions’, Proceedings of the Royal Society B, 289, available at .

Schleihauf, H., Sanford, E. M., Thompson, B. D., et al. (2025). ‘Chimpanzees Rationally Revise Their Beliefs’, Science, 390, pp. 521–26.

Wittgenstein, L. (1953). Philosophical Investigations, Blackwell.