In discussing Mahr & Csibra's (M&C's) observations about the role of episodic memory in grounding social commitments, we propose that episodic memory is especially useful for gauging cases of implicit commitment and cases in which the content of a commitment is indeterminate. We conclude with some thoughts about how commitment may relate to the evolution of episodic memory.

Mahr & Csibra (M&C) view autonoesis as being essential to episodic memories and construction as being essential to the process of episodic remembering. These views imply that episodic memory is systematically misleading, not because it often misinforms us about the past, but rather because it often misinforms us about how it informs us about the past.

Dissociative identity disorder is characterised by the presence in one individual of two or more alternative personality states (alters). For such individuals, the memory representation of a particular event can have full episodic, autonoetic status for one alter, while having the status of knowledge or even being inaccessible to a second alter. This phenomenon appears to create difficulties for a purely representational theory and is presented to Mahr & Csibra (M&C) for their consideration.

The distinctive feature of episodic memory is autonoesis, the feeling that one's awareness of particular past events is grounded in firsthand experience. Autonoesis guides us in sharing our experiences of past events, not by telling us when our credibility is at stake, but by telling us what others will find informative; it also supports the sense of an enduring self.

Mahr & Csibra (M&C) make a compelling case for a communicative function of episodic remembering, but a less compelling case that this is its primary function. Questions arise on whether confirming their predictions would support their account sufficiently, on the communicative function of preserving rich, nonbelieved memories, and on the epistemic benefits of developing false memories via the acceptance of misinformation.

Two arguments imply that Mahr & Csibra's (M&C's) functional theory is insufficient as an explanation of episodic memory: (1) The developmental course supports a different social cultural division of episodic and semantic memory, and (2) the existence of long-term autobiographical memory is not explained in the functional theory but can be seen in a broader cultural framework.

Improved control of agency is likely to be a prior and more important function of episodic memory than the epistemic-communicative role pinpointed by Mahr & Csibra (M&C). Taking the memory trace upon which scenario construction is based to be a stored internal model produced in past perceptual processing promises to provide a better account of autonoetic character than metarepresentational embedding.

Mahr & Csibra (M&C) fail to make the important distinction between why a trait originally evolved, why it was maintained over time, and what its current utility is. Here we point out that episodic memory may have originally evolved as a by-product of a general metarepresentational capacity, and that it may have current functions beyond the communicative domain.

Mahr & Csibra (M&C) claim that episodic remembering's autonoetic character serves as an indicator of epistemic authority. This proposal is difficult to reconcile with the existence of confabulation errors – where participants fabricate memories of experiences that never happened to them. Making confabulation errors damages one's epistemic authority, but these false memories have an autonoetic character.

Mahr & Csibra (M&C) include interesting ideas about the nature of memory from outside of the field of cognitive psychology and cognitive neuroscience. However, the target article's inaccurate claims about those fields limit its usefulness. I briefly review the most serious omissions and distortions of the literature by the target article, including its misrepresentation of event memory, and offer suggestions for forwarding the goal of understanding the communicative function of memory.

Mahr & Csibra's (M&C's) account of the communicative function of episodic memory relies more heavily on the case against episodic memory in nonhumans than their description suggests. Although the communicative function of episodic memory may be accurate as it pertains to human behaviour, we question whether Morgan's canon is a suitable foundation on which to build theories of supposedly human-specific traits.

According to Mahr & Csibra (M&C), the view that the constructive nature of episodic memory is related to its role in simulating future events has difficulty explaining why memory is often accurate. We hold this view, but disagree with their conclusion. Here we consider ideas and evidence regarding flexible recombination processes in episodic retrieval that accommodate both accuracy and distortion.

Mahr & Csibra (M&C) propose that episodic memory evolved to support epistemic authority in social communication. We argue for a more parsimonious interpretation whereby episodic memory subserves a broader preparatory function for both social and non-social behavior. We conclude by highlighting that functional accounts of episodic memory may need to consider the complex interrelations between self and subjective time.

Episodic memories are distinct from semantic memories in that they are epistemically generative and privileged. Whereas Mahr & Csibra (M&C) develop a metarepresentational account of epistemic vigilance, we propose an explanation that builds on our notion of scenario construction: The way an event of the past is presented in episodic memory recall explains the epistemic generativity and privilegedness of episodic memory.

In response to the commentaries, we clarify and defend our characterization of both the nature and function of episodic memory. Regarding the nature of episodic memory, we extend the distinction between event and episodic memory and discuss the relational role of episodic memory. We also address arguments against our characterization of autonoesis and argue that, while self-referential, it needs to be distinguished from an agentive notion of self. Regarding the function of episodic memory, we review arguments about the relation between future mental time travel and memory veridicality; clarify the relation between autonoesis, veridicality, and confidence; and finally discuss the role of episodic memory in diachronic commitments.

Does it make sense for people to hold one another responsible for what they do, as happens in countless social interactions every day? One of the most unsettling lessons from recent psychological research is that people are routinely mistaken about the origins of their behavior. Yet philosophical orthodoxy holds that the exercise of morally responsible agency typically requires accurate self-awareness. If the orthodoxy is right, and the psychology is to be believed, people characteristically fail to meet the standards of morally responsible agency, and we are faced with the possibility of skepticism about agency. Unlike many philosophers, I accept the unsettling lesson from psychology. I insist, however, that we are not driven to skepticism. Instead, we should reject the requirement of accurate self-awareness for morally responsible agency. In Talking to Our Selves I develop a dialogic theory, where the exercise of morally responsible agency emerges through a collaborative conversational process by which human beings, although afflicted with a remarkable degree of self-ignorance, are able to realize their values in their lives.

Nietzsche anticipates both the anti-reflective and the dialogical aspects of Doris's theory of agency. Nietzsche's doctrine of will to power presupposes that agency does not require reflection but emerges from interacting drives, affects, and emotions. Furthermore, Nietzsche identifies two channels through which dialogical processes of person-formation flow: sometimes a person announces what she is and meets with social acceptance of that claim; sometimes someone else announces what the person is, and she accepts the attribution.

We support John Doris's criticism of “reflectivism” but identify three shortcomings: (1) his neglect of humans' evolved predispositions and tendencies, (2) his failure to appreciate that identity and responsibility arise first from parsing our world ontologically, in a process we call “existential framing,” and (3) a potentially alarming implication of his “dialogic” model of identity formation: if identity is negotiated across diverse social situations, why isn't dissociative identity disorder more common?

Collaborativism is the view that moral reasoning is better when it is socially embedded. We propose that, when people take part in dialogic exchanges, they align in ways that open up novel avenues for sharing values and rationality criteria and, therefore, for exercising responsible agency. The hypothesis that collaborative interactions unfold through the alignment of minds and bodies helps articulate Doris's participatory approach to morality.

Parallels from visual processing support Doris's cognitive architecture underlying moral agency. Unconscious visual processes change with conscious reflection. The sparse and partial representations of vision, its illusions, and hallucinations echo biases in moral reasoning and behaviour. Traditionally, unconscious moral processes are developed by teaching and reflection. Modern neuroscience could bypass reflection and directly influence unconscious processes, creating new dangers.