Mahr & Csibra argue that the (alleged) self-referential and metacognitive (“autonoetic”) nature of episodic memory is a distinctively human adaptation subserving communication and claims to epistemic authority. But this argument is puzzling. The most that this communicative function requires is a robust disposition to describe one's memories in metacognitive terms. Whenever one has a conscious episodic event memory, which comprises perspectival modality-specific representations of an event felt as occurring some distance in the past, it will generally be an easy matter to report that one remembers it, and trivial, too, to report that one saw or heard (or whatever modality is appropriate) the event in question. Thus, the experiential mode of one's original experience of the event can be read directly off the content of the event memory itself. And the fact that it is a memory (as opposed to past-directed imagining) can generally be determined swiftly from the context (such as the question one has just been asked about the event, which evokes the memory) and/or from the speed, specificity, and vividness with which the memory emerges in consciousness. There is simply no need for an episodic event memory to possess self-referential or metacognitive content intrinsically. These contents can readily be computed at the time of reporting.
It might be argued that routine classification of one's episodic event memory as (say) a memory of seeing the event (in such a way that one experiences oneself as remembering having seen it) is to be expected as another instance of “thinking for speaking” (Slobin Reference Slobin, Gumperz and Levinson1996). Because one regularly needs to report on one's event memories in metacognitive terms in order to claim epistemic authority over the events in question, it makes sense that one should automatically conceptualize one's memories in such terms, in such a way that they appear in consciousness already with the right format for verbal report; or so it might be claimed. But other alleged instances of thinking for speaking have been thoroughly critiqued (Papafragou et al. Reference Papafragou, Massey and Gleitman2002; Reference Papafragou, Hulbert and Trueswell2008; Trueswell & Papafragou Reference Trueswell and Papafragou2010). There is no reason to believe that the present case should be any different. How one needs to report an event seems not to influence how one conceptualizes it; so why should the need for metacognitive reports of episodic memory influence the contents of those memories themselves?
It might more plausibly be argued that episodic event memories are routinely conceptualized in metacognitive terms to facilitate metacognitive self-management. One needs to keep one's future episodic imagining, one's past episodic imagining, and one's counterfactual imagining distinct from one's episodic remembering. Different functional roles – differing patterns of inference and decision making – will be warranted in each case. Hence, it might be adaptive for the category membership of such states to be built into their content during the construction process. One would thus experience oneself as imagining the future, as imagining counterfactually, or as remembering the past. This would then be thinking for thinking (or thinking for cognitive self-management) rather than thinking for speaking.
Even this thinking for thinking idea seems unnecessary, however. Episodic event memories will have a sense of pastness built into their content anyway (thus clearly distinguishing them from future imagining). And in most cases, the context in which the remembering occurs and such factors as immediacy, specificity, and vividness can themselves be the cues that trigger the appropriate forms of reasoning and decision making. We know that metamemory judgments depend on such cues (Mitchell & Johnson Reference Mitchell, Johnson, Tulving and Craik2000). So it is unclear why it must be the metacognitive judgment rather than the cues themselves that determine the state's functional role.
It seems that neither thinking for speaking nor thinking for thinking provides a good reason for believing that human episodic event memories should have an essentially autonoetic, self-referential character. As a result, there is no reason to think that when memories with such a character do occur, they constitute a special, distinct kind of memory; nor is there any reason to think that there is a difference of kind between the episodic memories of humans and those experienced by nonhuman animals. In both cases, episodic event memories will be apt to emerge in consciousness fully formed, involving feelings of (some degree of) pastness, together with modality-specific perspectival representations of the events in question. Sometimes among humans such memories might be classified and experienced in the moment as memories; but on other occasions, one might experience just the first-order contents in question. This will depend on contextual factors, relevance, and the needs of the moment, just as whether one experiences a plant as a bush or as an azalea can vary with context and relevance. The level up to which an episodic memory is conceptualized (having just first-order, or rather first-order-plus-metacognitive, content) is unlikely to be any more fixed than is the conceptualization of experience quite generally. And for what it is worth (not much; see Carruthers Reference Carruthers2011; Schwitzgebel Reference Schwitzgebel2011), I can report from my own case that my conscious episodic event memories are generally purely first order in nature. When I remember, I experience a set of perspectival modality-specific images of an event felt as occurring some distance in the past. But that I am remembering is infrequently a component of the experience itself.
Mahr & Csibra argue that the (alleged) self-referential and metacognitive (“autonoetic”) nature of episodic memory is a distinctively human adaptation subserving communication and claims to epistemic authority. But this argument is puzzling. The most that this communicative function requires is a robust disposition to describe one's memories in metacognitive terms. Whenever one has a conscious episodic event memory, which comprises perspectival modality-specific representations of an event felt as occurring some distance in the past, it will generally be an easy matter to report that one remembers it, and trivial, too, to report that one saw or heard (or whatever modality is appropriate) the event in question. Thus, the experiential mode of one's original experience of the event can be read directly off the content of the event memory itself. And the fact that it is a memory (as opposed to past-directed imagining) can generally be determined swiftly from the context (such as the question one has just been asked about the event, which evokes the memory) and/or from the speed, specificity, and vividness with which the memory emerges in consciousness. There is simply no need for an episodic event memory to possess self-referential or metacognitive content intrinsically. These contents can readily be computed at the time of reporting.
It might be argued that routine classification of one's episodic event memory as (say) a memory of seeing the event (in such a way that one experiences oneself as remembering having seen it) is to be expected as another instance of “thinking for speaking” (Slobin Reference Slobin, Gumperz and Levinson1996). Because one regularly needs to report on one's event memories in metacognitive terms in order to claim epistemic authority over the events in question, it makes sense that one should automatically conceptualize one's memories in such terms, in such a way that they appear in consciousness already with the right format for verbal report; or so it might be claimed. But other alleged instances of thinking for speaking have been thoroughly critiqued (Papafragou et al. Reference Papafragou, Massey and Gleitman2002; Reference Papafragou, Hulbert and Trueswell2008; Trueswell & Papafragou Reference Trueswell and Papafragou2010). There is no reason to believe that the present case should be any different. How one needs to report an event seems not to influence how one conceptualizes it; so why should the need for metacognitive reports of episodic memory influence the contents of those memories themselves?
It might more plausibly be argued that episodic event memories are routinely conceptualized in metacognitive terms to facilitate metacognitive self-management. One needs to keep one's future episodic imagining, one's past episodic imagining, and one's counterfactual imagining distinct from one's episodic remembering. Different functional roles – differing patterns of inference and decision making – will be warranted in each case. Hence, it might be adaptive for the category membership of such states to be built into their content during the construction process. One would thus experience oneself as imagining the future, as imagining counterfactually, or as remembering the past. This would then be thinking for thinking (or thinking for cognitive self-management) rather than thinking for speaking.
Even this thinking for thinking idea seems unnecessary, however. Episodic event memories will have a sense of pastness built into their content anyway (thus clearly distinguishing them from future imagining). And in most cases, the context in which the remembering occurs and such factors as immediacy, specificity, and vividness can themselves be the cues that trigger the appropriate forms of reasoning and decision making. We know that metamemory judgments depend on such cues (Mitchell & Johnson Reference Mitchell, Johnson, Tulving and Craik2000). So it is unclear why it must be the metacognitive judgment rather than the cues themselves that determine the state's functional role.
It seems that neither thinking for speaking nor thinking for thinking provides a good reason for believing that human episodic event memories should have an essentially autonoetic, self-referential character. As a result, there is no reason to think that when memories with such a character do occur, they constitute a special, distinct kind of memory; nor is there any reason to think that there is a difference of kind between the episodic memories of humans and those experienced by nonhuman animals. In both cases, episodic event memories will be apt to emerge in consciousness fully formed, involving feelings of (some degree of) pastness, together with modality-specific perspectival representations of the events in question. Sometimes among humans such memories might be classified and experienced in the moment as memories; but on other occasions, one might experience just the first-order contents in question. This will depend on contextual factors, relevance, and the needs of the moment, just as whether one experiences a plant as a bush or as an azalea can vary with context and relevance. The level up to which an episodic memory is conceptualized (having just first-order, or rather first-order-plus-metacognitive, content) is unlikely to be any more fixed than is the conceptualization of experience quite generally. And for what it is worth (not much; see Carruthers Reference Carruthers2011; Schwitzgebel Reference Schwitzgebel2011), I can report from my own case that my conscious episodic event memories are generally purely first order in nature. When I remember, I experience a set of perspectival modality-specific images of an event felt as occurring some distance in the past. But that I am remembering is infrequently a component of the experience itself.