Your browser is no longer supported

For the best possible experience using our website we recommend you upgrade to a newer version or another browser.

Your browser appears to have cookies disabled. For the best experience of this website, please enable cookies in your browser

We'll assume we have your consent to use cookies, for example so you won't need to log in each time you visit our site.
Learn more

How well do patients recall past experience of pain?

  • Comment

VOL: 103, ISSUE: 42, PAGE NO: 34-35

Rohini Terry, PhD, BSc; Catherine A. Niven, PhD, BSc, RGN; Eric E. Brodie, PhD, BAC, Psychol, AFBPsS

Rohini Terry is research fellow, department of psychology, University of Stirling; Catherine A. Niven is director, Nursing, Midwifery and Allied Health Professions Research Unit, University of Stirling; Eric E. Brodie is reader, department of psychology, Glasgow Caledonian University.

When patients have experienced pain, their decisions about future healthcare events are likely to be influenced by their recollections of previous painful health-related experiences. Patients are also often required to make judgements about the relative nature of their pain in order to indicate to health professionals if their condition is getting better or worse.
Rohini Terry, PhD, BSc; Catherine A. Niven, PhD, BSc, RGN; Eric E. Brodie, PhD, BAC, Psychol, AFBPsS
Rohini Terry is research fellow, department of psychology, University of Stirling; Catherine A. Niven is director, Nursing, Midwifery and Allied Health Professions Research Unit, University of Stirling; Eric E. Brodie is reader, department of psychology, Glasgow Caledonian University.

Abstract Terry, R. et al (2007) How well do patients recall past experience of pain?

BACKGROUND: Memory of past experiences of pain may affect patients’ decisions about future healthcare interventions, but methodological and ethical difficulties have hindered research into how people remember pain.

METHOD: A preliminary investigation highlighted limitations in inferring pain recall accuracy by simply comparing participants’ ratings of pain made while they were experiencing it with those made retrospectively. Memory for experimentally induced acute pain was then investigated using the ‘remember/know’ research paradigm. This allowed participants to indicate whether their recollections reflected clearly remembered, specific aspects of the pain, their earlier reports of the pain, or that they simply knew they had experienced pain.

RESULTS: Retrospective reports of pain intensity and quality were consistent with those made while the pain was experienced. Participants primarily reported remembering their pain when asked to recall it, and these recollections were not based on recollection of previously used pain descriptions.

DISCUSSION: Establishing the relative contribution of different types of memory will allow for a better understanding of how pain memories are constructed.

CONCLUSION: Better understanding of how pain memories are constructed will improve understanding of how attitudes and behaviours towards future pain.


When patients have experienced pain, their decisions about future healthcare events are likely to be influenced by their recollections of previous painful health-related experiences. Patients are also often required to make judgements about the relative nature of their pain in order to indicate to health professionals if their condition is getting better or worse.

Methodological and ethical problems make it difficult to study memory for pain, and a number of issues remain difficult to address. While previous research has suggested that memory for pain intensity might be quite accurate (Singer et al, 2001; Everts et al, 1999; Erskine et al, 1990), there is less consensus in the findings of studies investigating memory for more specific qualitative dimensions (Terry and Gijsbers, 2000; Brodie and Niven, 2000; Niven and Brodie, 1996; Beese and Morley, 1993; Roche and Gijsbers, 1986; Hunter et al, 1979), but these discrepancies may be due to differences in methodological and theoretical underpinnings, rather than differences in the accuracy of pain memory (Brodie and Niven, 2000).

The study of memory for pain is further complicated by the apparent similarities between reports of an actual experience of pain and estimates made by people with no personal experience of the pain (Brodie and Niven, 2000; Niven and Brodie, 1996). This suggests that ‘memory’ of a prior pain may actually reflect patients’ beliefs of what they think the pain ‘should’ have been like, or recollections of behavioural and emotional consequences of their experience of pain (for example, recollections of taking painkillers).

When investigating how people recall pain, clinicians therefore need to disentangle patients’ reports that reflect them clearly, consciously remembering an episode of personally experienced acute pain from those which reflect the fact that they simply know that an experience was painful. The distinction between the two types of awareness – remembering and knowing – has been drawn in the pain memory literature (Eich, 1993) but never empirically investigated.

It is also important to assess memory for pain in a way that allows participants to indicate whether their retrospective reports are based on recollections of the pain per se or previously made pain ratings. For example, if a patient rates an acutely painful experience as a ‘9’ and is later asked to recall this pain, it is important to know whether the patient is recalling the rating of ‘9’ or the intensity of the pain itself.

Preliminary studies

We first conducted two pilot studies to investigate:

  • Whether pain intensity is recalled more accurately than its qualitative aspects;
  • Whether it is possible to make appropriate estimates of a pain that has never been experienced. If this was found to be the case, it would call into question the extent to which retrospective reports can be taken to reflect ‘memory’ of pain, even if they closely match reports made while experiencing pain.

We recruited 74 participants aged 26–72 years, all of whom spoke fluent English. Twenty-four were patients undergoing day surgery for varicose vein removal. The remaining 50 were healthy adults from a non-clinical population, who were divided into two groups and asked to provide estimates of pain following vascular surgery, based on information provided in one of two leaflets about the surgery. None had recently undergone surgery or were expecting to in the foreseeable future, and none had ever undergone surgery for varicose vein removal.

Having obtained ethical approval, a researcher (RT) approached the group of patient participants at the day surgery unit after they had been admitted. They were asked to provide ratings of the intensity and quality of pain in the first 48 post-operative hours, prior to taking any analgesia.

These ‘actual pain’ ratings were made after the patients had been discharged from hospital, and returned to the researcher by post. Participants who returned the questionnaires were then asked to provide ‘retrospective pain’ ratings 4–6 weeks later by thinking back to the time they had completed the ‘actual pain’ questionnaire to recall any pain or discomfort they experienced specifically at that time.

In order to investigate the similarities between ratings made by patient participants and the estimates made by those without first-hand experience of the pain, the two groups of non-patient participants were required to provide estimates of post-operative pain following varicose vein surgery, based on the information provided in either a comprehensive information leaflet about varicose vein surgery (the information leaflet group) or a shorter, much less informative leaflet (the general leaflet group).

Both groups made two estimates of pain, the first (Time 1) immediately after reading their leaflet, the second (Time 2) 4–6 weeks later, without referring ba ck to the leaflet. It was explained clearly in writing that at Time 2 participants should think back to the time when they provided their last estimate and to base their second estimate on their recollections of that made at Time 1, without re-reading their leaflet. They were told: ‘There are no “right” or “wrong” answers, and the study is not trying to “test” your memory for information contained in the leaflets.’

The Short Form McGill Pain Questionnaire (SF-MPQ) (Melzack et al, 1985) was used to obtain ratings of pain quality and intensity at each assessment time. Like the original long version of the MPQ (Melzack, 1975),the SF-MPQ yields a rating purported to reflect pain intensity, the Pain Rating Index (PRI) and provides a description of the qualitative nature of the pain experience. A visual analogue scale (VAS), a 100mm horizontal line with the words ‘no discomfort’ and ‘worst possible discomfort’ at each end, was also used to obtain an estimate of pain intensity.

Actual and retrospective pain ratings provided by the patients were compared with the estimates of pain provided at Time 1 and Time 2 by the non-patients. Actual pain and retrospective ratings and non-patient estimates provided at Time 1 and Time 2 were compared within group to provide a measure of pain recall accuracy and rating consistency.

Differences between participant groups and pain rating times were investigated using analysis of variance (ANOVA). Kappa analysiswas used to assess the degree of similarity between each participant’s choice of SF-MPQ descriptors at each rating time. Kappa incorporates a correction for ‘chance’ agreement (the reselection of verbal descriptors) between the two times, and for descriptors which the participant has correctly not endorsed. It yields a value between zero, reflecting agreement no better than chance, and one, which reflects perfect agreement. Fleiss (1981) recommended that the level of agreement should be categorised as follows: Kappa 0.75–1.0 = excellent; 0.6–0.75 = good; 0.4–0.6 = fair; and <0.4 =="">

No significant differences were found between intensity ratings of actual experiences and retrospective ratings made by the patient participants and the estimates made by the healthy non-patient participants. However, Kappa analyses revealed inconsistencies in the choice of SF-MPQ descriptors used to express pain quality, with values of 0.25–0.36, reflecting a poor degree of consistency between descriptors chosen from the SF-MPQ. These findings are consistent with prior research (for example, Erskine et al, 1990.

We also wanted to investigate whether it is possible for people to make appropriate estimates of a pain they have never experienced – this would call into question the extent to which retrospective reports of pain can be taken to reflect ‘memory’ of that pain, even if they closely matched earlier ‘real-time’ ratings.

Our findings indicated that although both groups of non-patient participants overestimated the intensity of the pain, their estimates of its qualitative nature were similar to those of the patient participants. This highlights a major limitation in the memory for pain literature in that it does not tell us whether recollections of pain reflect memory of the experience, or are based on other non-experiential knowledge such as what the pain ‘should’ have been like.

This important distinction can be investigated using a research model widely used in cognitive psychology, known as the remember/know paradigm (Tulving, 1985). In the context of pain recollection, remembering would refer to the distinct conscious awareness of experiencing the pain, or ‘mental time travel’ (Tulving, 1985). However, this phenomenological experience should not be seen as sensory re-experiencing, or what Katz and Melzack (1990) termed ‘somatosensory memory’, which occurs only very.

Knowing about a prior pain experience, on the other hand, refers to a recollective experience which does not invoke any specific details and involves no conscious recollection of the experience (Tulving, 1985). It may consist of recollections of previously experienced episodes which are accompanied by a ‘feeling of knowing’ but are not consciously remembered, or of non-experiential semantic information, meaningful rules, or facts about the type of pain in question. The use of the remember/know paradigm should provide valuable insights into how we represent pain – as a piece of knowledge from the past (knowing) or as an experience that is recalled vividly (remembering).

A further difficulty faced by clinicians when interpreting patients’ retrospective pain ratings is deciphering whether they reflect memories of previous pain experiences, or previously expressed pain ratings (Clark and Bennett-Clark, 1993). The remember/know paradigm can also be used to examine this issue. More specifically, if participants report consciously recalling some aspect of the pain, using, for example, the MPQ descriptor ‘nagging’, they can be asked to judge whether they remember the nagging pain per se, or having previously chosen the descriptor ‘nagging’ to express it. Alternatively, both the pain and the descriptor may be remembered.

After the preliminary investigations, we used the remember/know paradigm in a controlled experimental setting to examine memory for acute pain. It was hypothesised that recollections would involve both remembering and knowing. The remember/know paradigm was also used to gauge the extent to which retrospective reports reflected a conscious recollection of the pain description as opposed to the pain sensation per se.


Research design

An experimental research design was used to control as far as possible for the potentially confounding variables inherent in clinical settings, such as anxiety and analgesic use, and to circumvent other methodological and ethical issues.

Acute pain was induced using a cold pressor (CP) test, a method widely used in pain research (Mitchell et al, 2004), which used a 150x285x155mm bath of water set at 5oC (+0.1oC) into which the participant places a hand and forearm, resulting in intense acute pain.

The MPQ was used to obtain qualitative ratings of CP pain in addition to the overall numerical indication of pain, the Pain Rating Index (MPQ-PRI). Participants were also required to complete the VAS used in the preliminary study at each rating time to describe pain intensity.


University ethical approval was granted and participants signed a consent form after receiving written and verbal information on the study. Ninety-seven university staff (41 men and 56 women) participated; the mean age for for male participants was 46.4 years (SD 10.8) and for female participants 39.9 (SD 9.7).

Participants were asked to immerse their non-dominant hand into the CP until they found the pain intolerable – they were asked not to use distraction techniques to keep the hand in for longer. When participants immersed their hand, the researcher waited approximately five seconds before starting to read the MPQ, from which they selected descriptors they felt appropriate to describe their pain.

Participants were told to remove their hand from the water if they found the pain intolerable and to replace it when they felt able to do so if the MPQ was still being read to them. When the whole of the MPQ had been read, they were told that they could remove their hand from the water and dry it and then to mark on the VAS the intensity of the CP pain experienced at its worst.

Recalling cold pressor pain

Part 2 of the study took place two weeks after Part 1, in a different room to reduce visual and situational cues. Participants were asked to think back to the CP test and try to recall the sensations experienced and to select appropriate MPQ descriptors to report their recollections as the questionnaire was read to them. They were then told they would be asked to make the ‘memory judgement’ about the MPQ descriptors they had just selected, which was to distinguish between remembering and knowing about their CP experiences. Until this point they wereunaware that they would be asked to make further judgements of their pain recollections.

Remember and know judgements

The difference between ‘remembering’ and ‘knowing’ about prior events and experiences was explained to participants (Conway et al, 1997; Rajaram, 1993; Tulving, 1985), then they were given the opportunity to describe their understanding of the two states of awareness. Additional examples of the differences were provided as necessary.Thus, participants were asked to indicate whether they consciously remembered specific sensations implied by the MPQ descriptor (a remember judgement). If they did not feel that they remembered the pain sensation they were asked to report that they simply ‘knew’ the descriptor chosen was appropriate to describe their recollections (a ‘know’ judgement). Participants could also report remembering that they previously selected an MPQ descriptorand whether or not this recollection was accompanied by a recollection of the pain sensation itself. Both remember sensation and descriptor judgements were assumed to reflect that the participant also knew that the pain descriptor was appropriate (Tulving, 1985).

Statistical analyses

Correlation analysis was used to compare numerical pain ratings (VAS and MPQ-PRI scores). Kappa was used to compare the specific MPQ descriptors and categories selected to provide a qualitative description of CP pain. The proportions of MPQ descriptors endorsed at recall as ‘remember sensation’ and ‘know’ were calculated.


Intensity ratings

Table 1. Mean ratings of pain using the VAS and MPQ

Actual pain

Retrospective ratings

Total-PRI (sd)

29.0 (13.1)

28.7 (13.0)

100mm VAS (sd)

52.7 (23.2)

52.7 (21.9)

Actual pain ratings and retrospective ratings were significantly correlated, VAS r = 0.83, p<0.01, mpq,="">r = 0.80 p<>

Qualitative ratings

The pain profile (Fig 1) illustrates pattern of MPQ descriptors selected while using the CP and two weeks later. The Kappa values reflecting the consistency of MPQ descriptor use at each assessment time was calculated at 0.45 (SD 0.18), reflecting a ‘fair’ degree of consistency. Kappa values reflecting the consistency of MPQ category use were ‘good’; 0.61 (SD 0.16).

Figure 1: Pain profiles showing the most frequently selected pain descriptors as chosen when using the CP and at two-week recall

Proportion of remembering and knowing

All participants felt that they rememberedat least some of the pain sensations reflected by the MPQ descriptors they had chosen, but not all endorsed MPQ descriptors at recall as either remember descriptor or know. Some 58% endorsed at least one as remember descriptor, while 74% endorsed at least one as know. Table 2 details the proportion of descriptors judged at recall as remember sensation, remember descriptor and know. Because participants could report that they remembered selecting the descriptor and remembered the sensation, the total number and proportion of remember sensation, remember descriptor and know judgements was greater than 1.0.

Table 2. Mean proportions of MPQ descriptors selected

Endorsed as:

Remember sensation

Remember descriptor


Mean proportions


0.73 (0.18)


0.20 (0.21)


0.23 (0.17)



In agreement with previous research (Singer et al, 2001; Everts et al, 1999; Beese and Morley, 1993), actual and recall ratings of pain intensity were closely correlated and numerical intensity ratings were reported two weeks later at the same level as when the pain was experienced. Conversely, Kappa values illustrated that concurrence between the use of qualitative MPQ descriptors was only moderate; and that participants did not always use the same descriptors at recall as while experiencing the CP pain. However, MPQ category use was significantly more consistent than MPQ descriptor use, with Kappa values reflecting a ‘good’ degree of agreement between actual and recall category selection. Melzack (1975) pointed out that the differences between the meanings of some MPQ descriptors within a category are subtle, so it is perhaps unreasonable to expect perfect retrospective recall. Discrepancies between actual and retrospective descriptor use may be due to limitations in the ability to verbalise the subjective experience of pain, rather than poor recollection. Perhaps when investigating the recall of the qualitative dimensions of pain, it is more appropriate to assess accuracy at the level of the MPQ categories, which reflect the defining properties of the pain, rather than at the level of the MPQ descriptors.

Almost three-quarters of the MPQ descriptors chosen at recall were selected because the participants felt they remembered the pain when using the CP, reflecting a clear, conscious recollection of the sensation. Previous research has suggested that memory for pain may be ‘more a matter of “knowing” than of “remembering”’(Eich, 1993) but we did not find this to be the case and it was possible to demonstrate that memories of acute CP pain were predominantly made up of a combination of clear, conscious recollections.

To a lesser extent, retrospective reports were based on simply knowing that certain types of pain would have been experienced when submersing a hand into ice-cold water. Less than a quarter of the MPQ descriptors were selected because participants simply found them appropriate (know judgements), without remembering anything specific about the pain implied by the descriptor, or having selected it previously. Whether these ‘know’ judgements reflect non-experiential knowledge of what a particular pain ‘should’ be like, or implicit memory or memories which are not verbally accessible, is unclear. But whatever memory systems are reflected in the know responses, these judgements may play a crucial role in the cognitive processing of pain events. For example, an individual may ‘know’ that certain medical procedures are painful and therefore avoid them (such as visits to the dentist) without actually ‘remembering’ specific episodes. Further research is required to examine the complex relationship between aspects of pain which are not consciously remembered but simply known to have occurred and behavioural responses to future pain.

From a clinical perspective, it is also important to decipher whether a patient feels they are remembering the nature of a pain or a previous description of it. Participants in this study endorsed only a small proportion of the ‘remember’ judgements as ‘remember descriptor’ (around 22% of descriptors selected at recall). That is, they did not tend to select a descriptor retrospectively because they remembered selecting it while using the CP. The majority of words picked retrospectively reflected their attempts to describe a remembered subjective sensory experience. Moreover, when they did indicate that a descriptor had been selected retrospectively because they remembered selecting it while in pain, they often reported that remembering the descriptor was accompanied by a conscious recollection of the sensation. These findings indicate that participants did not, by and large, attempt to remember their pain by recalling previous verbal ratings.

Limitations and future directions

The findings of this study are limited in that recall ratings were obtained only at one time point – two weeks after the CP test. A shift from remembering to knowing has been found to occur in other ‘real world’ situations where the proportions of ‘remember’ judgements decrease over time and know judgements increase (Dudukovic and Knowlton, 2005; Conway, 2001; Conway et al, 1997). Further research needs to investigate this remember-to-know shift when recalling pain experiences over different time periods and the myriad factors which may affect this process.


Using the remember/know paradigm, it has been possible to demonstrate that recollections of the intensity and quality of acute pain involves both remembering and knowing, with participants reporting remembering approximately three times more often than knowing. In addition, the paradigm has shown that it tends to be the pain which is remembered rather than previous reports of it. In conclusion, the use of the remember/know paradigm in future studies should yield data that is a closer approximation of ‘memory for pain’ than that provided by previous research. This could facilitate important advances in our understanding of how pain memories are constructed. The consequences of these memories in clinical settings, in terms of shaping attitudes and behaviours towards future pain, are yet to be examined.

Implications for practice

In agreement with previous research, the intensity of pain seems to be recalled accurately. The broad defining sensations of the experience also seem to be well recalled. However, as both individuals with and without personal experience of a particular pain can provide appropriate descriptions of it, the remember/know paradigm was used to investigate pain memory more directly. It was possible to show for the first time that pain memories are based primarily on clear, conscious recollections of pain sensations, rather than a recollection of previous pain ratings or simply knowing that a previous experience was painful.


Beese, A., Morley, S. (1993) Memory for acute pain experience is specifically inaccurate but generally reliable. Pain; 53: 2, 183–189.

Brodie, E.E.,Niven,C.A. (2000) Remembering an everyday pain: the role of knowledge and experience in the recall of the quality of dysmenorrhoea. Pain; 84: 1, 89–94.

Clark, W. C., Bennett-Clark, S. (1993) Remembrance of pains past? APS Journal; 2: 3, 195–200.

Conway, M.A. et al (1997). Changes in memory awareness during learning: The acquisition of knowledge by psychology undergraduates. Journal of Experimental Psychology: General; 126: 393–413.

Conway, M.A. (2001) Sensory-perceptual episodic memory and its context: autobiographical memory. Philosophical Transactions of the Royal Society of London, Series B, Biological Sciences; 356: 1413, 1375–1384.

Dudukovic, N.M., Knowlton, B.J. (2005) Remember-Know judgements and retrieval of contextual details. Acta Psychologica; 122: 2,160–173.

Eich, E. (1993) On the accuracy of memory for pain. APS Journal; 2: 3, 192–194.

Erskine, A. et al (1990) Memory for pain: a review. Pain 41, 255–265.

Everts, B. et al (1999) Pain recollection after chest pain of cardiac origin. Cardiology; 92: 2, 115–120

Fleiss, J.L. (1981) Statistical Methods for Rates and Proportions. New York: Wiley. [check – cited as 1985?]

Hunter, M. et al (1979) Memory for pain. Pain; 6: 35–46.

Katz, J., Melzack, R. (1990) Pain ‘memories’ in phantom limbs: review and clinical observations. Pain; 43: 3, 319–336.

Melzack, R. (1975) The McGill Pain Questionnaire: major properties and scoring methods. Pain; 1: 3, 277-299.

Melzack, R. et al (1985). The role of compensation in chronic pain: analysis using a new method of scoring the McGill Pain Questionnaire. Pain; 23: 2, 101–112.

Mitchell, L.A. et al (2004) Temperature and the cold pressor test. The Journal of Pain; 5: 4, 233–238.

NivenC.A., Brodie, E.E. (1996) Memory for labor pain: context and quality. Pain; 64: 2, 387-392.

Rajaram, S. (1993) Remembering and knowing; Two means of access to the personal past. Memory and Cognition; 21: 1, 89–102.

Roche, P., Gijsbers, K. (1986) A comparison of memory for induced ischemic pain and chronic rheumatoid pain. Pain; 25: 337–343.

Singer, A.J. (2001) Ability of patients to accurately recall the severity of acute painful events. Academic Emergency Medicine; 8: 3, 292–295.

Terry, R., Gijsbers, K. (2000) Memory for the quantitative and qualitative aspects of labour pain: a preliminary study. Journal of Infant and Reproductive Psychology; 18: 2, 143–151.

Tulving, E. (1985) Memory and consciousness. Canadian Psychologist; 26: 1–12.

  • Comment

Have your say

You must sign in to make a comment

Please remember that the submission of any material is governed by our Terms and Conditions and by submitting material you confirm your agreement to these Terms and Conditions. Links may be included in your comments but HTML is not permitted.