Mahr & Csibra's (M&C's) evolutionary hypothesis is that episodic memory has the proper function of signaling and evaluating the veracity of reports about non-occurrent events. This evolutionary hypothesis entails a causal hypothesis: that episodic memory makes a very significant contribution to our ability to tell convincing stories about the past, replete with context and detail, and to our ability to probe those details to test whether the stories others tell us are true. We find the evolutionary hypothesis less compelling and fertile than the causal hypothesis.

We reject M&C's argument in favor of the evolutionary hypothesis. Their argument assumes that among the many things episodic memory allows us to do, just one can be singled out as the dominant cause of its having been selected, that is, that the various hypotheses about episodic memory's proper function are mutually exclusive. We see it as altogether more likely that episodic memory evolved under multiple selection pressures: for witnessing (M&C), for future thought (Schacter et al. Reference Schacter, Addis and Buckner2007), for gossip (Keven Reference Keven2016a), and for counterfactual reasoning (De Brigard Reference De Brigard2014a). The precise features of episodic memory likely reflect a compromise between these selective pressures and sharp design constraints on the set of permissible variations to existing functional parts. It is therefore exceptionally risky to infer the actual causal history of a trait such as episodic memory from its current form. It is comparatively safe to study what episodic memory does here and now, for example, to test the causal hypothesis.

The causal hypothesis is speculative. Its value lies primarily in its ability to suggest new and productive research questions. We consider three such novelties: for neuropsychology, comparative psychology, and philosophy.

From a neuropsychological perspective, individuals with episodic amnesia might be a valuable source of evidence for evaluating the causal hypothesis. Episodic memory and future-oriented episodic thought are typically operationalized by asking participants to tell stories about past, future, or counterfactual events and then counting how many internal (i.e., story-relevant) details they provide. And it is well known that individuals with acquired amnesia generate fewer details in their narratives than do neurotypical humans, supporting the causal hypothesis (Hassabis et al. Reference Hassabis, Kumaran and Maguire2007a; Reference Hassabis, Kumaran, Vann and Maguire2007b). That said, it is also known that individuals with perinatal hippocampal damage never develop the capacity for episodic memory and have nonetheless been shown to generate copious detail if properly cued (Hurley et al. Reference Hurley, Maguire and Vargha-Khadem2011; Mullally et al. Reference Mullally, Hassabis and Maguire2012), though their stories have been reported to lack spatiotemporal coherence. We wonder how M&C would explain this finding. Concerning the second, evaluative, component of the causal hypothesis, it is currently unknown whether individuals with amnesia are generally perceived to be less credible than neurotypical individuals and, perhaps more interestingly, whether individuals with episodic deficits sort veridical from non-veridical testimony comparably to controls. Might an episodic memory deficit, perhaps even from birth, be expected to influence how well one can assess the firsthand testimony of others? The causal hypothesis thus opens new questions about the role episodic simulation might play in the second-order evaluation of other people's narratives.

Consider, next, comparative psychology. The causal hypothesis might allow investigators to assess whether nonhuman creatures and preverbal children are truly capable of episodic thought. Controversy in this area stems in part from the fact that there is no accepted behavioral indicator of autonoetic experience. Experimental paradigms used to assess “episodic-like” memory in scrub jays, for example, establish only that they know what they cached, where, and when. Clayton and Dickinson (Reference Clayton and Dickinson1998) are forced to call this memorial capacity “episodic-like” precisely because the method cannot establish that the birds reconstruct the caching event or know that they previously cached the seeds. However, if the causal hypothesis is correct, the existence in a group of organisms of a complex set of practices for enforcing veridical and detailed information delivery about non-occurrent events counts as evidence that those organisms have some autonoetic capacity. If, for example, children began to enforce veridical reporting of non-occurrent events prior to the acquisition of verbal language, that would provide some evidence that they keep track of the details of non-occurrent events and hold beliefs about them to epistemic standards. One might view the observation of rich communicative practices as a very high bar for discovering episodic memory. On the other hand, this standard might reveal something unique about episodic memory in humans.

Finally, we turn to philosophy. In many arenas of human social life (friendship, law, journalism, politics, and science), we treat firsthand experience as a special source of knowledge. The witness to the atrocity is treated as knowing things that others cannot; the witness knows them “directly” rather than via testimony. Witnesses hold trump cards: They can speak with authority about certain aspects of the event, whereas non-witnesses cannot. The epistemic value of first-person experience is grounded in a set of social practices (of recognizing, regimenting, and enforcing the telling of truths about the past), and these practices are possible only if creatures in that social group have the capacity to reconstruct events and attribute them to their personal pasts. Even granting the many well-known failures of episodic memory as a reliable source of information about the past (Schacter Reference Schacter2001), the fact that our cognitive systems contain such a mechanism makes it possible for creatures like us to engage in the practices that constitute and sustain the privileging of firsthand experience. The causal hypothesis, in short, points the way to a deeper understanding of how the normative epistemic privilege of first-person experience is made possible by the cognitive mechanisms we possess.

In this case, our understanding of the space of causes (i.e., of the causal mechanisms that populate our neurocognitive apparatus) is linked to our understanding of the space of reasons (i.e., of how we justify our beliefs and actions) through a set of social practices that both create and sustain the norms constitutive of the epistemic privilege of the witness (the witness trump card). In our view, this philosophical thought is the core fruit of M&C's highly original target article.