We applaud Barzykowski and Moulin's (B&M) advocacy for a comprehensive account of memory retrieval. As they acknowledge, voluntary memory retrieval has long served as the default. The key terms and processes of retrieval are characterized on the basis of their operation in voluntary contexts. The success of a unified account is thus dependent on how easily concepts that were developed to account for voluntary memory phenomena transfer to involuntary memory phenomena.

Here, we focus specifically on the concept of cues. All memory is cued, at least in the sense that memory is the recurrence, re-experiencing, or reconstructive representation of prior information, events, or activities. For an item retained in memory, the potential for reactivation is more or less constant. A memory coming to mind now, rather than at some other point in time (or not at all), is explained at least in part by some internal or external thing serving as a cue for its reactivation.

As causes of memory retrieval, cues precede their effects. B&M's comprehensive retrieval model is built around this, using cues as the starting point for all retrieval processing (as depicted in their Fig. 1). While this is true of both voluntary and involuntary memory – that is, that they both start from cues – the details of this process are importantly different across these two forms of memory.

In voluntary remembering, there is a clearly identified cue which is used to prompt a clearly identified target. Within such a framework, there is a clear starting point and a clear end, as well as success conditions, making it possible to measure the retrieval speed, perceived fluency, and accuracy/error of the remembering attempt. Importantly, in such cases, the cue can be identified prior to and independently of the memory it activates. The causal process from the cue to retrieval is thus extended and consciously available to the would-be rememberer.

The same is not true for cases of involuntary remembering. Involuntary memories and déjà vu are cued, in the sense that all memories are, but it is not possible to identify the cue prior to the involuntary activation of a memory. B&M characterize IAMs and déjà vu as cases of “spontaneous processing of cues.” This is, in effect, an acknowledgement that, in cases of involuntary retrieval, a cue can only ever be identified in retrospect. The retrieval process begins with a cue, but the rememberer is not aware of this process until the memory is activated. Moreover, there is no sense in asking whether the retrieved memory is the one targeted by the cue; there is no opportunity for it to be otherwise.

There is also no sense to be made of the idea of a cue failing to activate a memory during involuntary retrieval. If nothing is activated, then there is no cue – and also no memory. This is depicted in the leftmost column of the retrieval flowchart B&M provide in Figure 1. It is, however, strange to depict it at all. It is somewhat misleading to characterize proceeding straight down that column – answering no to the intentionality of retrieval, access to content, and a feeling of familiarity – as resulting in “no memory.” It is true that there is no memory in such cases, but that seems to be because there is no retrieval process at all. Nothing happens.

If this is the case, then it raises further problems for how we understand the depiction of involuntary memory and déjà vu in B&M's model. In terms of what is consciously accessible, the starting point in cases of involuntary memory and déjà vu is not the cue, but the activation of a memory. The modification required for involuntary memory may be minimal, as these cases involve consciously accessing content. How B&M will accommodate this for déjà vu is less clear.

The complicated relationship between involuntary memories and their cues is further illustrated by cases where voluntary and involuntary retrieval are intermixed. B&M combine voluntary and involuntary memory into a single model, but they keep each form of retrieval distinct. Phenomena such as chaining (e.g., Mace, Reference Mace2006; Pilotti, Chodorow, & Tan, Reference Pilotti, Chodorow and Tan2004) suggest that a complete model should also account for interactions between these forms of retrieval. Chaining begins with voluntary memory, when the attempt to retrieve the targeted memory activates other memories inadvertently. An instance of voluntary remembering is thus chained to one or more instances of involuntary remembering.

Such a case nicely illustrates the distinct role cues play in voluntary and involuntary memory. In a case of chained remembering, the voluntary memory and the involuntary memory share a cue. Their relation to this cue, however, is different. The voluntary memory is targeted by that cue; the involuntary memory is not. The cue is understood as a cue for the voluntary memory from the outset. It only becomes a cue for the involuntary memory once that memory is activated. If the involuntary memory had not been activated, then it would not be a cue for that memory. Not so for the voluntary memory: Even if the retrieval attempt failed, the cue remains a cue.

B&M are right to push for greater attention to involuntary memory processes and a comprehensive account of retrieval. As this brief exploration of cues demonstrates, however, we think there is further work yet to be done.