ReviewMeasuring recollection and familiarity: Improving the remember/know procedure
Highlights
► Recollection and familiarity are thought to both contribute to recognition. ► They are commonly measured with the remember/know (RK) procedure. ► We review evidence that the RK procedure is meaningful and consider alternatives. ► We suggest ways to use the procedure as accurately as possible. ► The RK procedure, when used with care, is a valid and important method.
Introduction
Much research in memory focuses on our experience of remembering; whether we use recollection or familiarity. Many different terms are used, often interchangeably, although important distinctions need to be made between task definitions and terms that describe kinds of memory. Some terms help differentiate between the type of memory task being used (e.g., recall versus recognition) whereas others help differentiate between the kind of memory that can support it (e.g., episodic versus semantic memory or recollection versus familiarity). Here we focus on assessing the contributions of recollection and familiarity to recognition, using the remember/know procedure. There is much theoretical debate about the cognitive and neural interactions underlying recollection and familiarity, therefore having solid, reliable measurement methods is critical for interpretation.
The remember/know procedure is popular, widely used but has also been criticised. There are many behavioural variants, some more suitable than others. Before discussing these methodological issues and since our viewpoint is that the RK procedure can be a valuable method, we include a brief review of the evidence supporting the idea that RK judgements meaningfully reflect recollection and familiarity. We then discuss alternative models of RK performance. We believe that the RK procedure and particularly modifications of it are often more useful than other experimental methods assessing familiarity and recollection, and these will be considered. The role of recollection and familiarity in memory needs to be related to memory testing in general, with the distinction between recall and recognition being an important starting point. The complex way in which the terms described in this paper relate to each other is shown in Fig. 1.
In experimental memory tasks, participants often study lists of items or associations before being tested with a recall or recognition test. In recall tests, some form of cueing is provided, which can vary in specificity. Cues can be previously studied stimuli in their entirety, specific cues for particular items (usually referred to as ‘cued’ recall), or cues that are general to encoding as a whole (usually referred to as ‘free’ recall). Recalled information can be episodic and/or semantic (see Tulving (1972) for episodic/semantic distinction).
In ‘episodic’ recognition tests, participants judge whether stimuli are old or new. These tests are episodic because they intend participants to identify whether a stimulus was encountered in a previously experienced context. In ‘semantic’ recognition tests, participants make true or false decisions without needing to retrieve any specific previous study context. However, participants may recognise an ‘episodic’ stimulus without remembering anything about a specific study context or make a true/false decision using memory for a specific context in which a stimulus occurred. This means that semantic and episodic tasks sometimes use contributions from both types of memory.
Various dissociations in recall and recognition performance have been seen, particularly in certain patient groups who show disproportionate recall compared to recognition deficits (Aggleton and Shaw, 1996, Bauman and Murray, 1968, O’Carroll, 1995). There is, however, a problem in interpreting recall/recognition dissociations; more than one type of memory is believed to contribute to recognition performance. In particular, according to dual process theories (Mandler, 1980, Yonelinas, 2002), recognition performance depends on recollection (a form of cued recall) as well as familiarity. Historically, familiarity has often been equated with recognition, particularly in early computational accounts of memory (e.g., Gillund & Shiffrin, 1984), but it is important to differentiate between the memory process (familiarity) and the memory test being used (recognition). They should not be confused. As shown in Fig. 1, both recollection and familiarity contribute to recognition performance. Different patterns of breakdown of recollection and familiarity are reported in ageing and in different patient groups. This makes it theoretically important to measure their separate contributions to recognition and determine whether their processing and neural bases are at least partially distinct. The relative contributions and accuracy of these kinds of memory can vary markedly across different recognition tasks.
Recollection occurs when a stimulus cues the recall of details linked to it in a previous encounter, e.g., bumping into someone and recalling that you last saw them at the Christmas party doing the conga. The memory comprises associations between a studied stimulus and details from their previous encounter (i.e., the study context). Familiarity is experienced as a feeling that you have had prior exposure to a stimulus without the recall of any associated details from prior exposure(s). A common example is meeting someone who you recognise without being able to recall anything about them, which can be an awkward social experience (i.e., the ‘butcher on the bus phenomenon’; Mandler, 1980). The critical aspect of the recollection experience is bringing to mind information that is not present in front of you, i.e., it involves cued recall. Recollection and familiarity, therefore, differ with respect to both memory content (amount of ‘context’) and underlying processes (recall or no recall).
Recollection is by definition episodic; it involves memory for details from specific study contexts. However, very similar forms of cued recall are semantic because they involve memory for factual information only. When/if cued recall eventually occurs in a ‘butcher on the bus’ situation, this provides a good example of a possible confusion between episodic and semantic memory. Recalling the butcher’s name and/or occupation, without any recall of a particular episode where you met before, would be semantic cued recall rather than episodic recollection. To be classified as episodic recollection, you would need to recall details from a previous occasion on which you saw the butcher. Semantic and episodic memory recall systems can often interact, although how their underlying processing differs remains largely unexplored. It seems reasonable and probable to expect them to depend, at least in part, on similar underlying cognitive processes/neural substrates.
If familiarity exists for stimulus–context associations, then it is episodic. If there is familiarity for factual information (i.e., feeling that a fact is true), then it is semantic. There may also be familiarity for some combination of semantic and episodic information, or information not easily classified as either, such as occurs when a face feels familiar. Above-threshold familiarity involves a conscious feeling of memory so cannot be equated with ‘unconscious’ memory for a stimulus (c.f. Joordens, Wilson, Spalek, & Paré, 2010), even if that feeling is very weak. Measures of familiarity and priming, therefore, need to be carefully separated (Voss & Paller, 2007). Also, as discussed later, the dominant view (Atkinson and Juola, 1973, Jacoby and Dallas, 1981, Mandler, 1980, Yonelinas, 1994) is that familiarity is based on a signal that varies along a dimension of strength in a normally distributed fashion. When recognition of studied stimuli fails, the familiarity signal may often be stronger than the signal from unstudied stimuli, but still fall below the recognition criterion strength. This ceases to be the case at the limit when signal strength of studied and unstudied stimuli is the same.
In an episodic recognition task, both familiarity and recollection can contribute to making decisions about whether a stimulus is old or new. This raises several theoretical questions. First, how and why do the relative contributions of familiarity and recollection differ across tasks? Second, are the psychological (e.g., use of strategic, effortful processing versus automatic nature of processing) and brain mechanisms that underlie the two kinds of memory the same or partially distinct? There has been a particular focus on whether the neural bases differ within the medial temporal lobes (MTLs; Brown and Aggleton, 2001, Eichenbaum et al., 2007, Squire et al., 2007) and to a lesser degree, in the parietal lobe (Simons and Mayes, 2008, Vilberg and Rugg, 2009). Third, how do recollection and familiarity combine for recognition decisions (Cohen et al., 2008, Higham and Vokey, 2004, Parks and Yonelinas, 2007, Schumacher et al., 2010, Wixted, 2007, Wixted and Mickes, 2010)? In order to examine these theoretical issues, recollection and familiarity must be accurately and separately measured. Here we focus on the remember/know (RK) procedure, which is discussed in detail in Section 2.
Despite the use of the RK procedure in the development of theories of recognition memory, one critical aspect of the task is often ignored; the practical details of how the task is carried out. An overlooked area centres on how researchers can be confident that their participants are following their instructions. It seems timely to evaluate the advantages and disadvantages of this procedure and to start to work towards a consensus on what should be ‘best practice’.
In this review, we draw attention to weaknesses in the method that must be addressed for RK data to be interpretable in terms of understanding the component processes of recognition memory separately and ideally, also how they function in combination. In particular, we summarise the work indicating that varying the details of RK instructions can affect participants’ responses. Additionally, we recommend that more effort is made to check that participants understand and follow the instructions. Other methods for adapting the RK procedure to achieve different theoretical aims are also considered.
Section snippets
The remember/know procedure
The remember/know (RK) procedure (Tulving, 1985) is the most widely used method to investigate dual process theories of recognition memory. Participants are asked to decide whether a previously studied item is ‘remembered’ from the study episode or whether they just ‘know’ that it is old. Novel or ‘new’ distractor stimuli should be identified as such. When using the RK procedure to investigate recollection and familiarity, ‘remember’ (R) responses are taken as recollections and ‘know’ (K)
Other methods of estimating recollection and familiarity
A significant advantage of the RK procedure is the ability to differentiate between recollection and familiarity on a trial-by-trial basis. Task dissociation methods, such as speed-accuracy trade-off tasks and tasks comparing recall and recognition are inexact, making process estimation methods more attractive (see Yonelinas (2002) for a full discussion of task-dissociation and process estimation methods). These process estimation methods include receiver operating characteristic (ROC) curves
Evidence of separate neural bases for R and K responses
There has been much work looking at the neural bases of recollection and familiarity, using diverse methods across different species. When looked at as a whole, multiple studies have concluded that there are separable brain regions/processing underlying the different experiences of memory and, by implication, the different kinds of memory that are believed to underlie them (Diana et al., 2006, Eacott and Easton, 2007, Rugg and Curran, 2007, Spaniol et al., 2009, Yonelinas, 2002). There is,
RK and confidence
The strength of a memory must be related to the underlying biology. With long-term memory, strength is presumably a function of the extent and/or type of synaptic changes within the neural circuit supporting a memory. The extent of such changes is likely to relate poorly to the duration of a memory because this will be strongly influenced by post-consolidation factors; however, unfortunately no other direct storage-related measure currently exists. Instead, indirect proxy measures related to
False recollection
A problematic issue for dual process accounts of recognition memory is R false alarms (Wixted & Mickes, 2010). We know that false recall and R false alarms can occur in patients who confabulate (Ciaramelli & Ghetti, 2007) and in false memory paradigms (e.g., Roediger & McDermott, 1995). It is important to determine whether these false memories relate more to recollection that is incorrect or to recollection that is correct not diagnostic of stimulus oldness. For example, in a false memory
Recollection and familiarity distributions
Familiarity is usually seen as normally distributed along a continuum of strength, with recollection as a threshold process (e.g., Yonelinas, 1994). Another approach views both recollection and familiarity as normally distributed overlapping distributions (Wixted, 2007). We often experience subjective differences in confidence about familiar information. This also occurs with recalled information, such as being more or less sure we are correct after recalling someone’s name. This means that
Using the remember/know procedure
The recommendation to use the RK procedure over other tasks is contingent on the specific way the procedure is implemented. The remainder of this paper is concerned with these details of the methods, centring on the current variation in the literature. Regardless of whether our specific ideas on details of instructions or checking procedures are completely agreed upon, we strongly argue for the need to report RK procedures in a more detailed and consistent way.
The RK procedure differs from
Conclusions
Across multiple studies, with different populations and with imaging work, there is considerable evidence that RK experiments are measuring recollection and familiarity. This suggests that the RK procedure is measuring something useful and meaningful and, when used carefully, is not just assessing recognition confidence. There are no alternative methods that are clearly better at accessing feelings of recollection and familiarity on a trial-by-trial basis. A clear area that requires resolution
Acknowledgments
We would like to thank Steve Lindsay, Matthew Howard, Alex Kafkas and Anne Schmechtig for their feedback on a previous version of this manuscript.
References (184)
- et al.
Amnesia and recognition memory: A re-analysis of psychometric data
Neuropsychologia
(1996) - et al.
Sparing of the familiarity component of recognition memory in a patient with hippocampal pathology
Neuropsychologia
(2005) - et al.
The role of the temporal lobes in recognizing visuospatial materials: Remembering versus knowing
Brain and Cognition
(1997) - et al.
Remembering and knowing in context
Journal of Memory and Language
(2003) - et al.
Recollective and nonrecollective recall
Journal of Memory and Language
(2010) - et al.
Re-examining dissociations between remembering and knowing: Binary judgments vs. independent ratings
Journal of Memory and Language
(2011) - et al.
How many response options? A study of remember–know testing procedures
Acta Psychologica
(2010) - et al.
What are confabulators’ memories made of? A study of subjective and objective measures of recollection in confabulation
Neuropsychologia
(2007) Effects of attention and confidence on the hypothesized ERP correlates of recollection and familiarity
Neuropsychologia
(2004)- et al.
False recognition after a right frontal lobe infarction: Memory for general and specific information
Neuropsychologia
(1997)
True and phantom recollection: An fMRI investigation of similar and distinct neural correlates and connectivity
NeuroImage
Imaging recollection and familiarity in the medial temporal lobe: A three-component model
Trends in Cognitive Sciences
Remember–know judgements and retrieval on contextual details
Acta Psychologica
Changes in familiarity and recollection across the lifespan: An ERP perspective
Brain Research
Experiences of remembering, knowing, and guessing
Consciousness and Cognition
On interpreting the relationship between remember–know judgments and confidence: The role of instructions
Consciousness and Cognition
Recollection-based recognition: Word frequency effects
Journal of Memory and Language
Retrieval dynamics of recognition and frequency judgments: Evidence for separate processes of familiarity and recall
Journal of Memory and Language
A process dissociation framework: Separating automatic from intentional uses of memory
Journal of Memory and Language
Intra- and inter-item associations doubly dissociate the electrophysiological correlates of familiarity and recollection
Neuron
Recollection, familiarity, and cortical reinstatement: A multivoxel pattern analysis
Neuron
Turning the process-dissociation procedure inside-out: A new technique for understanding the relation between conscious and unconscious influences
Consciousness and Cognition
The contribution of recollection and familiarity to yes–no and forced-choice recognition tests in healthy subjects and amnesics
Neuropsychologia
The relationship between remembering and knowing: A cognitive neuroscience perspective
Acta Psychologica
Creating illusions of familiarity in a cued recall remember/know paradigm
Journal of Memory and Language
Examining the neural basis of episodic memory: ERP evidence that faces are recollected differently from names
Neuropsychologia
Recognition memory: A review of the critical findings and an integrated theory for relating them
Cognitive Psychology
Recognition memory and the human hippocampus
Neuron
Associative memory and the medial temporal lobes
Trends in Cognitive Sciences
Increased reading speed for words and pronounceable non-words: Evidence of preserved priming in amnesics
Cortex
On the validity of remember–know judgments: Evidence from think aloud protocols
Consciousness and Cognition
The influence of instructions and terminology on the accuracy of remember–know judgments
Consciousness and Cognition
Aging reduces veridical remembering but increases false remembering: Neuropsychological test correlates of remember–know judgments
Neuropsychologia
Explorations of familiarity produced by words with specific combinations of letters
European Journal of Cognitive Psychology
Factors influencing the speed and accuracy of word recognition
Preserved recognition in a case of developmental amnesia: Implications for the acquisition of semantic memory?
Journal of Cognitive Neuroscience
Preserved visual recognition memory in an amnesic patient with hippocampal lesions
Hippocampus
CA1 neurons in the human hippocampus are critical for autobiographical memory, mental time travel, and autonoetic consciousness
Proceedings of the National Academy of Sciences of the United States of America
The contribution of recollection and familiarity to recognition memory: A study of the effects of test format and aging
Neuropsychology
The contribution of familiarity to within- and between-domain associative recognition memory: Use of a modified remember/know procedure
European Journal of Cognitive Psychology
Recognition versus recall in schizophrenia
Canadian Journal of Psychology
The episodic nature of involuntary autobiographical memories
Memory and Cognition
Episodic memory and remembering in adults with Asperger syndrome
Journal of Autism and Developmental Disorders
Impaired familiarity with preserved recollection after anterior temporal-lobe resection that spares the hippocampus
Proceedings of the National Academy of Sciences
Conjoint recognition
Psychological Review
Recognition memory: What are the roles of the perirhinal cortex and hippocampus?
Nature Reviews Neuroscience
Recognition memory: Material, processes, and substrates
Hippocampus
Evaluating models of remember–know judgments: Complexity, mimicry, and discriminability
Psychonomic Bulletin and Review
Components of recognition memory: Dissociable cognitive processes or just differences in representational complexity?
Hippocampus
Cited by (116)
The metacognitive abilities of narcissists: Individual differences between grandiose and vulnerable subtypes
2024, Personality and Individual DifferencesAccelerated Long-Term Forgetting: Prolonged Delayed Recognition as Sensitive Measurement for Different Profiles of Long-Term Memory and Metacognitive Confidence in Stroke Patients
2022, Journal of the International Neuropsychological SocietyA conditional judgment procedure for probing evaluative conditioning effects in the absence of feelings of remembering
2024, Behavior Research Methods
- 1
These authors contributed equally to this work.