Blakey, KH., Rafetseder, E., Melis, G., Veit, A., Amelung, K., Freudensprung, F., Kovacs, K., & Virányi, Z. (2024). Non-verbal rationality? 2-year-old children, dogs and pigs show unselective responses to unreliability, but to different degreesPsyArXiv. ​ 


Some philosophers argue that reflection, the ability to assess one’s reasons for beliefs, is the defining feature of rational thinking. Yet, by tying reflective thinking to language, they struggle to account for apparently rational behaviour in minimally verbal human infants and altogether exclude non-human animals. To assess capacities for a basic form of reflective thinking that does not require language, we investigated whether 2-year-old children, dogs, or pigs could acquire and respond to undermining defeaters. In an object-search task, subjects observed different informants act on one of two screens: one informant’s actions reliably indicated the reward location, while the other’s actions were independent of the reward location. The informants switched to using new actions twice during a sequence of repeated trials, which put subjects in the position to make an inference about the reliability of the informants based on the informativeness of their actions. Neither 2-year-olds nor animals responded differently to the Reliable and Unreliable informants. There was a reduction in subjects’ willingness to follow the indications of the informants in later trials, which was stronger in 2-year-olds than animals. While these results provide no evidence that 2-year-olds, dogs, or pigs made an inference about the reliability of each informant across the actions, reduced willingness to follow the informants’ indication suggests that they may be responding to either uncertainty or an undermining defeater like <the informants are unreliable>.

Book chapter

Melis, G. Forthcoming Rationality and reflection in human and non-human animals. In Stovall, P. and Koreň, L. (eds.), Why and How We Give and Ask for Reasons, New York: Oxford University Press.


I challenge the view that reflection and language are essential to responsiveness to reasons from the standpoint of current analytic epistemology. After an introduction, in sections 2-4, I illustrate how the study of the cognitive abilities of very young children suggests that language and reflection are not necessary for responsiveness to reasons. In sections 5-6, I sketch a proposal in which both linguistic and non-linguistic subjects have access to the same broad rational abilities, if not to the same extent.

Journal article

Melis, G. & Monsó, S. Are Humans the Only Rational Animals? The Philosophical Quarterly, (2023).


While growing empirical evidence suggests a continuity between human and non-human psychology, many philosophers still think that only humans can act and form beliefs rationally. In this paper, we challenge this claim. We first clarify the notion of rationality. We then focus on the rationality of beliefs and argue that, in the relevant sense, humans are not the only rational animals. We do so by first distinguishing between unreflective and reflective responsiveness to epistemic reasons in belief formation and revision. We argue that unreflective responsiveness is clearly within the reach of many animals. We then defend that a key demonstration of reflective responsiveness would be the ability to respond to undermining defeaters. We end by presenting some empirical evidence that suggests that some animal species are capable of processing these defeaters, which would entail that even by the strictest standards, humans are not the only rational animals.

Journal article

Melis, G. Normative Defeaters and the Alleged Impossibility of Mere Animal Knowledge for Reflective Subjects. Philosophia (2023).


Investigating how subjects’ knowledge is structured and acquired is central to understanding the relation between the cognitive achievements of adult humans and those of very young children and non-human animals. One important distinction is the one between reflective and unreflective knowledge. The former is supported by an evaluation of one’s state of knowledge, and adult humans exhibit it when they think something like “I know that Andy Murray plays today because I checked the BBC website.” The latter is not supported by an evaluation of the relevant state, and it is often ascribed to young children and non-human animals. For example, a two year-old may know that her mummy is in the room with her because she sees her mummy, and yet be incapable of formulating a thought like “I know that mummy is here because I see her.”

According to philosophers Sanford Goldberg and Jonathan Matheson, mere unreflective knowledge—that is, unreflective knowledge that is not thereby also reflective knowledge—is impossible for reflective subjects. Goldberg and Matheson’s view may support a dualism about knowledge which, if correct, would frustrate attempts to provide a unified account of knowledge-attributions to human adults, very young children, and non-human animals.

This article discusses Goldberg’s and Matheson’s proposal, outlines some of the ways in which it is insightful, and argues that it fails to establish the impossibility of mere unreflective knowledge for reflective subjects because it neglects the inherent temporal dimension of knowledge acquisition. It is argued that reflective subjects can have plenty of mere unreflective knowledge. The paper also highlights connections between the debates on the relation between unreflective and reflective knowledge, propositional and doxastic justification, and the theory of epistemic defeat. The final section advocates an interdisciplinary approach to investigate the relationship between unreflective and reflective cognitive achievements.

Open science involves sharing research methods and results throughout the life cycle of a project. As a group, we are keen to establish and promote as many open science practises as possible, and are committed to open science principles.

We have created an Open Science Framework (OSF) page for the project. For each study, we will create an OSF project including (when available) the preregistration, publicly available data and analysis code, and when possible, a copy of the materials used in our experiments.

OSF | ARED Project

Theme by the University of Stirling