Remembering and Knowing

Remembering and knowing are two subjective states of awareness associated with memory.

Remembering entails mental time travel that intimately engages one’s sense of self.

Knowing includes the general sense of familiarity we have about more abstract knowledge. Knowing also includes awareness of events that we have personally experienced when we are aware of those events as facts, without reliving them mentally.

Throughout this entry the terms remembering and knowing are used only to refer to these conscious states and to the responses subjects make when reporting them.

General Background

The idea that remembering and knowing could be studied in the memory laboratory was suggested by Endel Tulving (1985), who proposed that the two states of awareness reflect autonoetic and noetic consciousness, two types of consciousness that respectively characterize episodic and semantic memory systems (Tulving, 1983, 1989; Wheeler, Stuss, & Tulving, 1997).

He reported illustrative experiments in which subjects were instructed to report their states of awareness at the time they recalled or recognized words they had previously encountered in a study list.

If they remembered what they experienced at the time they encountered the word—something they thought of at that time—they made a remember response. If they were aware they had encountered the word in the study list but did not remember anything they experienced at that time, they made a know response.

The results indicated that participants could quite easily distinguish between experiences of remembering and knowing. Though Tulving (1985) used free-call, cued-recall, and recognition tests, it was recognition memory that became the most commonly used remember/know paradigm, not least because recognition memory is most likely to be associated with experiences of knowing, as well as remembering, especially when recognition is accompanied only by feelings of familiarity.

Moreover, the two states of awareness captured by remember and know responses seemed at the time additionally relevant to dual-component theories of recognition memory, which held that recognition could be accomplished by either one of two independent processes, recollection and familiarity (e.g., Mandler, 1980).

The major premise underlying the use of remember and know responses is that the subjective states of awareness they measure cannot be reliably inferred from more conventional measures of performance.

For example, proportion correct may be equivalent in two different experimental conditions, or in two different subject groups, yet proportions of remembering and knowing may differ.

One cannot tell what subjects experience mentally from purely objective measures of their performance. If one wants to be able to take into account subjective awareness of memory, there is no alternative to the use of subjective reports. This does not, of course, mean that such reports have to be accepted at face value, since they can obviously be confabulated (see e.g., Dalla Barba, 1993). They have to be interpreted carefully, in conjunction with other evidence.

Remember and know responses are not intended as introspective measures of any underlying hypothetical constructs, such as memory systems or processes. Their use also differs from classical introspection in that all that is required is that subjects distinguish between kinds of mental experiences, rather than report the details of those experiences.

Importantly too, however, when subjects are additionally asked to describe their experiences in more detail, the results confirm that remember responses reflect awareness of what was experienced when a word was encountered in a study list and know responses do not, but reflected only awareness of recent but unremembered encounters that are attributed to the study list.

Major Findings

Studies that followed Tulving (1985) discovered various experimental manipulations that dissociate remembering from knowing. Many independent variables were found to influence remembering but not knowing, particularly variables that engage more conceptual and more elaborative processing, like level of processing, and generating versus reading at study (Gardiner, Java, & Richardson-Klavehn, 1996a).

Other variables, though not nearly so many, were found to affect knowing but not remembering, particularly variables that engage perceptual processing.

These variables include the presentation of a test word being preceded by an identical, masked test prime, compared with its being preceded by a masked but unrelated test prime, and same versus different study and test modalities following a highly perceptual orienting task (Gregg & Gardiner, 1994).

Some variables were found to have opposite effects on remembering and knowing. These variables include studyng nonwords compared to words. Nonwords led to increase knowing and less remembering (Gardiner & Java, 1990).

Massed versus spaced repetition of items within a study list was found to have a similar effect. Massed repetition led to more know responses and fewer remember responses than spaced repetition (Parkin & Russo, 1993).

Gardiner, Gawlik, and Richardson-Klavehn (1994) found a pattern of results that is analogous to showing a double dissociation between remembering and knowing within a single experiment.

This experiment used an item-by-item directed forgetting paradigm to functionally manipulate the relative amounts of elaborative and maintenance rehearsal. It was assumed that elaborative rehearsal would be focused on to-be-learned, rather than on to-be-forgotten, items.

Maintenance rehearsal was manipulated by varying the delay between the presentation of the item and the subsequent cue that designated it as to-be-learned or to-be-forgotten.

Because it is not in the interests of subjects to rehearse items elaboratively until they are told whether the items are to be learned or to be forgotten, it was assumed also that the longer the delay, the longer the period of maintenance rehearsal.

The results bore out these assumptions. Lengthening cue delay increased know but not remember responses. But item designation influenced remember but not know responses, with more remembering for the to-be-learned items. These results are summarized in Table X-1.