The Episodic self: the self as modelled in individual past events

The Episodic self is a feature that emerges from the combination of self-modelling and conscious memory recall. As an emergent feature, the Episodic self is neither directly learned nor directly innate; it is what becomes possible when there is an interaction between two other features that are, themselves, learned or innate (Pomerantz and Cragin 2015). The emergent feature of the Episodic self seems to be particularly interesting, because one of the interacting features is learned and the other is innate: the combination of the capacity to self-model (learned) and the capacity to remember events in the past (innate) creates the possibility of modelling a self in a remembered past event. Instead of the event being passively visceral – the emotions of the event are remembered as emotions – it becomes actively visceral – the emotions of the event are remembered as my own emotions. The Episodic self is, therefore, more than just an episodic memory, and it is equally as real as any self-model.

An Episodic self is not a memory of a past self-model; it is a current representation constructed from the social-self evidence currently available. When we remember our self, we do not remember our self-model as it was when the memory was laid down; rather, we construct a current self-model to represent our previous self. Giorgio Marchetti (2014) says this is because we are prone to three ‘sins’ of memory: we forget or mitigate the visceral emotions that were actually generated by the event, making our emotional memory of the event unreliable; we distort our memories by remembering the events themselves incorrectly, thus rendering our procedural memory of the event unreliable; and we over-emphasise some aspects of the event while under-emphasising others, thus pathologising our memory of the event. To use a von Neumann computer metaphor, recalling a memory is not just accessing a fixed memory-image like a file on a hard drive; it involves copying the memory-image into working memory, adjusting the memory and then writing it back as a new image (Schiller and Phelps 2011). But memory is not just accidentally inherently fallible; it is important that it be so, so that each time I recall the memory I can model the experiences in the past event as my experiences in relation to my current self-model.

Yet another feature makes an episodic self-model unreliable as a model of my previous self: it is composed of more than my own memory of the event. The sharing of social models is more effective if the contexts and evidence of those models are also shared; so any memory I have of a past event is overlaid with the memories of others about that event – and every viewpoint of the event is different. What I believe is my memory of the event, seldom is; instead it is, like my models of my self and others, an amalgam of viewpoints and opinions. The Episodic self is a memory of a self-model that was originally generated from the models of me offered by others, and when recalled it is then edited by my current model of me and by more models of me offered by others.

If the Episodic self is just a type of self-model, and self-modelling is an outcome of shared social calculus, and sharing social calculus seems to be limited to humans, then is it possible for non-humans to have Episodic selves? The simple answer would seem to be no; but, because the Episodic self is an emergent feature, the actual answer is more complex. One of the two components of an Episodic self is the capacity to recall past events, which Endel Tulving (2005) calls noetic memory, contrasting it with autonoetic memory (the capacity to recall past events that include the past self’s own perspective). Tulving takes the view that autonoetic memory is available only to humans. In his description, autonoetic memory seems similar to episodic memory, having both recall and a self-perspective. Tulving’s autonoetic self-perspective is that of a past self; but is that a true self-perspective, and is that past self really available to the current self as a self-model?

Some researchers disagree with Tulving’s view. Fabbro et al. (2015) propose that a capacity for autonoetic memory is likely to be present in non-human brains, because the neurologically complex brain areas associated with human selfhood have correlates in those non-human brains. However, it remains to be demonstrated that the correlate areas function in the same capacity in both brains. In contrast, Robert Numan (2015) differentiates non-human from human episodic memory by describing it as ‘episodic-like’. This is a more cautious approach that does not necessarily require the generation of an Episodic self. It also gives us a convenient way to label the difference between the autonoetic episodic selfhood of humans and the otherwise episodic memory that many mammals do appear to possess.

If episodic selfhood, like Tulving’s autonoetic memory, is an innate capacity in humans, then we have a problem explaining how it evolved. If, however, it emerges from the modelled self plus noetic memory, and the modelled self is an outcome of sharing social calculus, then we can say that episodic selfhood is a synthesis of pre-existing cognitive systems. It is a learned trick, a way of creating a third-person social calculus model that happens to represent the self. Autonoetic memory is not a species-difference requiring its own evolutionary explanation; it is just noetic memory plus a trick.