If humans are biologically engineered to appraise the world accurately, how can we explain misbeliefs? After asking this question, McKay & Dennett (M&D) analyse various misbeliefs. Those resulting from a breakdown in the system, and those that are by-products, do not threaten the claim of adaptiveness of the belief system. Positive illusions are the only bona fide example of misbeliefs. We shall integrate this account by first making a case for the adaptiveness of some mistakes in the conception of the physical world, and by discussing the possibility of a general egocentric bias in generating positive illusions.

The grand aim of Naive physics (NP) is to fully describe common beliefs about the physical world. Naive physics can be traced back to Gestalt psychologists such as Köhler, and to the seminal work by Lipmann and Bogen (Reference Lipmann and Bogen1923). The term is also used in artificial intelligence and robotics (Hayes Reference Hayes and Michie1978). Despite its grand aim, interest in NP has focused on the discovery that people make some systematic mistakes about everyday phenomena. Examples include judgements about the pendulum motion (Bozzi Reference Bozzi1958); predictions of motion of an object in terms of direction, path of motion, and acceleration (Hecht & Bertamini Reference Hecht and Bertamini2000; McCloskey et al. Reference McCloskey, Caramazza and Green1980); and predictions about what is visible in a mirror (Bertamini & Parks Reference Bertamini and Parks2005). In the case of the pendulum, people consider as “natural” a movement that is actually artificially contrived. We can be sure that some mistakes are not cultural whims because they match scientific theories of the past (i.e., Aristotelian mechanics). NP beliefs are not necessarily approximations or simplified representations of the physical world (Cavanagh Reference Cavanagh2005). In some cases the implied physics is complex, for instance, when subjects deem as correct cast shadows that require light to bend around corners or to be projected from physically impossible locations (Casati Reference Casati2008).

Even if these mistakes are the manifestation of (implicit) mental models (McCloskey Reference McCloskey1983), where do these models come from? Typically NP beliefs are resilient and non-revisable, thus pointing to some modular underlying mechanism. Some NP beliefs are grounded on evidence provided by the visual system. The belief that a pendulum looks unnatural when it moves, for example, originates from how people perceive motion (Bozzi Reference Bozzi1958; Pittenger Reference Pittenger1989). Aspects of how people reason are also important, as exemplified by the reliance on prototypes of actions (Yates et al. Reference Yates, Bessman, Dunne, Jertson, Sly and Wendelboe1988) and heuristics (Proffitt Reference Proffitt, Wilson and Keil1999). Mistaken beliefs that originate from properties of perceptual or reasoning mechanisms could be classified as evolutionary by-products. On the other hand, one can ask the question of why these as opposed to other by-products occur. System limitations should also be considered from an evolutionary standpoint. For example, if waitresses make larger mistakes than housewives in the water-level task (the orientation of water in a tilted glass) this may be because the glass as a frame of reference is more important to them in their job than it is to other people (Hecht & Proffitt Reference Hecht and Proffitt1995). This may seem paradoxical but it suggests that attention to a local frame of reference, which is crucial for a task, makes it harder to learn about more abstract frames of reference. Context is, therefore, critical here. At least some NP beliefs, we surmise, are examples of systematically mistaken adaptive beliefs. In spite of their wrongness they provide contextually useful representations.

We are not claiming that each specific NP belief is an adaptation. Our perceptual system and our thoughts may lead us to them as a response to a situation. This brings us to the second point of our commentary.

Adaptiveness itself is hard to assess. Veridicality is not sufficient as a criterion. Just like percepts, most beliefs are prima facie veridical (they do not interfere with our interactions with the world) but compliance with logic or the laws of physics is not what they (beliefs as well as percepts) have evolved towards. An adapted organism is one that has accumulated characteristics that maximise fitness, not knowledge per se. Positive illusions are adaptive because they lead people to engage in adaptive behaviours. Whatever the mechanism, positive views of one's medical condition and of one's ability to influence it lead to increased health. Quite possibly the effects are not directly in terms of guiding deliberation and choice, rather they are ancillary effects, such as triggering emotional adjustments and immune reactions. The evidence about biased responses concerning the self is vast, and controversial. It spans items as diverse as: self-serving biases and positive illusions (Taylor & Brown Reference Taylor and Brown1994b), implicit egotism (Pelham et al. Reference Pelham, Carvallo and Jones2005), narcissism (Nuttin Reference Nuttin1985), self-enhancement (Sedikides & Gregg Reference Sedikides and Gregg2008), and self-resemblance and trust (DeBruine Reference DeBruine2002), among others.

But are these beliefs specific adaptations or are they facets of a powerful but unspecific underlying mechanism, which we may call “looking after number one”? We think the jury is still out. If specific beliefs originate from specific adaptations, then it should be possible to find not only examples of “positive” illusions about oneself, but also of “negative” illusions about oneself that are, under different circumstances, adaptive. We would, therefore, need an example of a trait that is both generally perceived as positive (e.g., height) and yet such that people tend to see themselves as lacking because the resulting underestimation has a specific adaptive effect. If, on the contrary, we only have examples of overestimations (i.e., errors in the direction perceived as positive) then the most economical hypothesis is that they are all related, and originate from the same generic bias in favour of the self. Another problem with the idea that specific beliefs are specific adaptations is the fact that biases in favour of the self exist also for neutral or non-beneficial aspects. For instance, preferences are influenced by presence in their formulation of the first letter of the name of the person expressing the preference (Nuttin Reference Nuttin1985); compliance with a request increases when someone is told that they share a birthday with the requester (Burger et al. Reference Burger, Messian, Patel, del Prado and Anderson2004); and people overestimate the size of their own head (more than other people's heads) (Bianchi et al. Reference Bianchi, Savardi and Bertamini2008). It is unclear what the benefits are for these effects, and it seems more likely that they all originate from a generic (and adaptive) egocentric bias.