Does collaboration with an intimate partner support memory performance? An exploratory case series of people with epilepsy or acquired brain injury

2.1 Participants

Participants were recruited through a community brain injury service in Australia. Patients were invited to participate if they had a diagnosis of either ABI or epilepsy, had attended the local community brain injury clinic or had undergone a neuropsychological assessment at the first author’s private practice, and had a current partner. Exclusion criteria were relationship duration under one year, time since injury less than 3 months, and language impairment that would impact on their ability to complete the study tasks. Nine participants (4 female) with epilepsy (n = 2) or various types of ABI (n = 7; traumatic brain injury n = 1, stroke n = 3, tumour resection n = 1, hypoxic brain injury due to cardiac arrest n = 2) and their partners (total 18 participants) completed the study. Table 1 provides a summary of participant demographics including the main findings of their initial neuropsychological assessments performed for clinical purposes.

The Californian Verbal Learning Test – Second Edition (CVLT-II). The CVLT-II (Delis, Kramer, Kaplan, & Ober, 2000) is a verbal learning and memory task in which participants study and recall a list of words. The experimenter administered either the standard or alternate form of the CVLT-II and read aloud the 16 items to participants at the rate of one word every two seconds. After each of three presentations of the list, the experimenter instructed participants to immediately recall aloud as many words as they could remember within a time limit of three minutes. After the three presentation-test cycles (Tests 1-3), there was a 20-minute delay before participants were asked to recall the words again, without hearing them re-presented (long-delay free recall). For simplicity, we analyse and report Test 1 and the long-delay free recall for the CVLT task. We calculated the total number of words correctly recalled during Test 1 and the long-delay free recall of the CVLT task; we considered a word correct if it corresponded with the studied list or was a close approximation (e.g., singular/plural errors were considered correct).

Mutual Friends Task. We developed a personal semantic recall task in which the experimenter asked participants to recall the names of as many mutual friends or acquaintances (shared with their partner) as possible within a three-minute time limit. Specifically, the experimenters instructed participants: “I would like you to tell me all of the mutual friends or acquaintances both you and your partner know. Please only tell me people for whom you know both their first name and last name. Also, please do not include family members in your list. Try to make your list as long as possible. You will have three minutes to do this task”. If recall appeared blocked, the experimenter prompted participants to continue trying to think of additional items until three minutes had elapsed. We calculated the total number of friends correctly recalled on the Mutual Friends task; we considered a friend correct if it included both a first name and a surname and was generated within the time limit for that task.

First Meeting Task. Participants were asked to recall in detail the autobiographical event of meeting their partner for the first time, within a time limit of three minutes. Specifically, the experimenter instructed participants, “Think about the time when you first met your partner. I want you to describe this event to me, and make your description as detailed as possible, including what happened and where, as well as both of your reactions and emotions and anything else you can remember”. We coded the transcripts from the First Meeting autobiographical memory task for Episodic Specific details by tallying the number of details recalled that consisted of unique information relating to a specific episodic event; that is, details that referred to a unique, single event that happened on a particular day.

Personal Assessment of Intimacy in Relationships (PAIR). The PAIR scale (Schaefer & Olson, 1981) is a measure of couple intimacy consisting of 36-items across 5 sub-scales: Emotional Intimacy (6 items), Social Intimacy (6 items), Sexual Intimacy (6 items), Intellectual Intimacy (6 items), Recreational Intimacy (6 items) and a Conventionality Scale (6 items). Participants rate how much a given statement applies to their current romantic relationship on a 5-point Likert scale (1 = strongly disagree, 5 = strongly agree). Results were scored out of a possible maximum of 180, with higher scores indicating greater assessment of relationship intimacy.

Everyday Memory Questionnaire (EMQ). The EMQ (Sunderland, Harris, & Baddeley, 1983) is a self-reported measure of memory failures in everyday life consisting of 35-items across 5 sub-scales: Speech (13 items), Reading and Writing (4 items), Faces and Places (6 items), Actions (6 items), and Learning New Things (6 items). Participants rate how frequently each memory failure item has occurred over the past month on a 5-point Likert scale (for sub-scales 1-4, 0 = never, 4 = several times a day; for sub-scale 5, 0 = never, 4 = on every occasion). Each patient completed the EMQ in relation to themselves, while the healthy participants completed the EMQ in relation to their partner with an ABI or epilepsy. Results were scored out of a possible maximum of 140, with higher scores indicating greater frequency of everyday memory failures.

2.3 Case descriptions

Case 1 was a 22-year-old woman who sustained a severe traumatic brain injury (with neuroimaging evidence of bi-frontal and bi-temporal damage) in a motor vehicle accident 7 months prior to her participation in the study. She had been in a relationship with her boyfriend (‘high school sweethearts’) for 5 years with a period of separation for several months just prior to and after her injury. Their relationship intimacy ratings on the PAIR were well matched (see Table 2). Her opinion of her own memory functioning was fairly consistent with her boyfriend’s opinion, who rated her as having marginally more difficulties on the EMQ (see Table 3).

Case 2 was a 28-year-old woman who had a right parietal haemorrhage secondary to a ruptured aneurysm 3 months prior to her participation in the study. She had been married to her husband for 10 years. Their relationship intimacy ratings on the PAIR were inconsistent, with the husband rating significantly higher (see Table 2). Her husband’s and her own opinion of her memory function were well matched on the EMQ (see Table 3).

Case 3 was a 51-year-old man who had a left cerebellar stroke 11 months prior to his participation in the study. He had been married to his wife for 30 years. Their relationship intimacy ratings on the PAIR were well matched (see Table 2). His wife rated him as having slightly fewer difficulties than his self-rating on the EMQ (see Table 3).

Case 4 was a 27-year-old woman who underwent a tumour (craniopharyngioma at base of brain) resection 4 months prior to her participation in the study. She had been married to her husband for 9 years. Their relationship intimacy ratings were well matched on the PAIR (see Table 2). Her opinion of her memory functioning was inconsistent with her husband, who rated her as having fewer difficulties than her self-rating (see Table 3).

Case 5 was a 68-year-old man who sustained a hypoxic brain injury during a cardiac arrest approximately 11 years prior to his participation in the study. He had been married for 40 years. Their relationship intimacy ratings on the PAIR were highly inconsistent, with his rating much higher than his wife’s (see Table 2). His opinion of his memory function was inconsistent with his wife who rated his difficulties as much greater compared with his self-rating on theEMQ (Table 3).

Case 6 was a 37-year-old man who had a stroke (intraventricular haemorrhage of unknown cause) approximately one year prior to his participation in the study. He had been married for 10 years. Their relationship intimacy ratings on the PAIR were highly inconsistent with her rating much lower than his (see Table 2). His opinion of his memory function was inconsistent with his wife who rated his difficulties as much greater compared with his self rating on the EMQ (see Table 3).

Case 7 was a 53-year-old man who sustained a hypoxic brain injury during a cardiac arrest 10 months prior to his participation in the study. He had been in a relationship for 33 years. Their relationship intimacy ratings on the PAIR were highly inconsistent with her rating much lower than his (see Table 2). His opinion of his memory function was inconsistent with his wife’s, who reported his difficulties as much greater on the EMQ compared with his self-rating (see Table 3).

Case 8 was a 62-year-old man who had a diagnosis of temporal lobe epilepsy at 14 years of age that was revised to generalised idiopathic epilepsy in adulthood. He had been married for 30 years. Their relationship intimacy ratings on the PAIR were relatively well matched (see Table 2). His opinion of his memory function was somewhat inconsistent with his wife’s, who reported his difficulties as greater than his self-rating on the EMQ (see Table 3).

Case 9 was a 72-year-old woman with recent onset left temporal lobe epilepsy. She had been married 50 years. Their relationship intimacy ratings on the PAIR were relatively well matched (see Table 2). Her opinion of her own memory function was consistent with her husband’s, with their ratings on the EMQ well matched (see Table 3).

In summary, participants varied substantially in age, relationship duration and characteristics, and injury variables including nature and degree of cognitive impairment. We were interested in how these different profiles would impact on their individual and collaborative remembering.

3.1 Session 1

In Session 1, patients attended the brain injury service or the private practice office of the first author and were tested alone. Participants were first asked to recall autobiographical memories cued by a self-selected song and photograph, the results of which we do not report here. Participants were then asked to complete the CVLT task, followed by the Mutual Friends task and then the First Meeting task. At the end of Session 1 patients completed the PAIR.

3.2 Session 2

One week after the initial individual recall session, patients returned together with their partners and participated in a collaborative recall session. The experimenter administered the CVLT task using the alternate list to Session 1 and the same procedure of four recall tests but with an instruction for participants to work together to jointly recall the words. The experimenters did not instruct participants on how to collaborate or how to resolve disagreements, except that they should “work together to help each other to recall as many items as possible”. Next, participants completed the Mutual Friends task with the same instructions as in Session 1 except that couples should “work together to tell me all of the mutual friends or acquaintances that you both know”. We limited participants to three minutes for each of these joint recall tasks. If recall appeared blocked, the experimenter prompted participants to continue trying to think of additional items until three minutes had elapsed. As in Session 1, the experimenters asked participants to jointly recall in detail a number of autobiographical memories, including the memory of their first meeting, for which they were given three minutes. The instructions were the same as in Session 1, except that they were asked to “work together to recall as many details as possible”. At the end of Session 2, both patients and partners completed the EMQ and PAIR.

3.3 Coding and scoring

Two trained research assistants transcribed all sessions in full from the audio recordings. We scored the CVLT and Mutual Friends memory tasks for the total number of items (words, friends) recalled and the individual contribution of the patient during collaborative recall. Two independent raters coded for the amount of Episodic Specific detail during the First Meeting tasks from Session 1 and 2; their inter-rater reliability was r = .84. We retained the ratings of the first coder for analysis following discussion of any disagreements.

On the word list task, recall scores of the couple together were higher than for the patient alone across both initial learning and delayed recall, except when the patient was effectively at ceiling, in which case there was no room for significant improvement (i.e., Case 1 and 2 delayed recall score, see Table 4). In regard to the contribution of the participant with ABI or epilepsy during collaboration, with the exception of Cases 2 and 9 who showed marginally better Test 1 performance during collaboration compared with their individual performance, in all other cases, the patients contributed less during collaboration for both initial learning and delayed recall performance than when they performed the task alone (see Table 4). Therefore, collaboration typically resulted in decreased performance of the participant with ABI or epilepsy, and the improved overall score during collaboration was due to the contribution of their healthy partner.

4.2 Personal semantic memory

On the Mutual Friends task, performance (total score) of the couple together was higher than for the patient recalling alone in all cases (see Table 5). When we examined the contribution of the patients to the score during collaboration, three patients (Cases 1, 3 and 8) showed improved performance, recalling more names when collaborating with their partner than when they did the task alone. The remainder (6/9) contributed less names during collaboration compared with when they did the task alone (see Table 5). Of note, some patients generated new names during collaborating which they had not produced when doing the task alone. Therefore, there was evidence that in some cases, collaborating with a partner enabled patients to recall more names overall (Cases 1, 3 and 8), or to recall new items not previously recalled when doing the task alone (e.g., Cases 1, 3 and 9). Nevertheless, the majority of patients contributed less names overall when collaborating on this task.

4.3 Autobiographical memory

On the First Meeting autobiographical memory task, the majority of couples (7/9) recalled more episodic details when they collaborated together than the patient recalled alone (see Table 6). In two cases, however, collaboration actually decreased the overall score (Cases 4 and 5), in that the couple together provided fewer episodic details than the patient had been able to recall alone. When we examined the contribution of the patients during collaboration, in the majority of cases (6/9) they contributed fewer episodic details to the collaboration than they had been able to recall alone (see Table 6). In one couple (Case 3) the contribution was the same, and in two couples (Cases 6 and 7) collaboration improved patients’ individual performance. In summary, collaboration typically improved overall performance of the autobiographical memory task, but this was due to the healthy partner’s contribution, as in the verbal learning and recall task, while the patient’s contribution was typically less during collaboration than when they completed the task alone.

4.4 Interaction and scaffolding processes

There was variability in scaffolding styles and task performances across and within couples, indicating the contextual sensitivity of the scaffolding phenomenon. Nevertheless, we found that patients tended to recall less with their spouse than they were able to recall alone. Why did this reduction in contribution occur and why was it more marked in some couples than others? Analysis of the transcripts provides insights into the patterns of interaction that drove this overall tendency for patients to recall less when they collaborated with their spouse. We identified two distinct styles in the way that patients and their partners approached the task of collaborating together.

The first, we term “survival scaffolding”. This was characterised by healthy partners promptly ‘taking over’ the recall task, while the patients’ contributions tended to be repetitions and acknowledgments of their partner. This interaction style was particularly evident in Couples 5, 6 and 7, as the following transcript examples illustrate. The dialogue below shows Couple 6’s joint recall on the CVLT long-delay recall task:

The patient contributes the first item but then pauses, prompting their partner to take over, and their partner repeats the item ‘cricket’ without acknowledging that the patient had in fact already recalled it. Subsequently, the patient contributes the new item ‘grasshopper’, but is uncertain, repeating the word several times and seeking confirmation from their partner. Finally, the patient contributes ‘cricket’ again, but can’t recall whether he or his partner have already said it. Similarly, during Couple 5’s joint recall on the Mutual Friends task1

In the example above the healthy partner begins the recall, with the patient providing acknowledgements but not contributing new items. Later, when the patient doesn’t recognise the recalled names ‘Melissa and David Berger’, their partner provides a justification, “We haven’t seen them for a while”, before continuing with the task.

During Couple 5’s recall on the First Meeting task, the partner similarly begins recall and contributes most of the unique information:

In the example above, the partner provides an opportunity for the patient to contribute, but their response does not provide any new information, instead a vague indication that the story is complete (“progressed from there”). Later, the partner seeks confirmation of their recall with the question, “would that be right?”, but this is met by a simple acknowledgment without the patient providing any further detail.

When Couple 7 attempt the Mutual Friends task together, the healthy partner similarly directs the recall:

In this example, the patient’s early attempt to contribute the recall item “Graham” is lost while his wife simultaneously recalls friends’ names. When she later prompts him to continue with the task (“Keep going!”) he is hesitant and indicates that there are no more names to recall. His further contributions are limited to simple acknowledgements and repetition of his wife’s recall, as seen in Cases 5 and 6.

Interestingly, Couples 5, 6 and 7 all showed similar patterns on the EMQ and PAIR measures. On the EMQ, these three couples showed asymmetric results with the partners rating the patients as having significantly more everyday memory failures than the patients themselves (Table 3). On the PAIR, these three couples also showed highly asymmetric results with the patients reporting much higher relationship intimacy than their partner’s assessment of their relationship intimacy (Table 2). Therefore, it appears that mismatched ratings of memory function and intimacy between couples may be associated with this particular ‘survival scaffolding’ style in which the healthy partner takes primary responsibility for remembering.

This can be contrasted with a second style, which we term “stability scaffolding”. This was characterised by more symmetrical contributions to recall, and more frequent and strategic cueing by the healthy participants of their partner with a cognitive impairment. This interaction style was particularly evident in Couples 1, 2 and 9, as the following transcript examples illustrate. The dialogue below shows Couple 1’s joint recall on Test 1 of the CVLT task:

In the example above, the healthy partner begins the task by prompting the patient to contribute the items that they can remember first, and only contributes additional items after the patient pauses. In Couple 1’s joint recall on the Mutual Friends task, the partner similarly prompts the patient shortly after they begin recall:

The partner suggests the strategy, “Let’s go through your friends”, choosing a category relevant to the patient and therefore perhaps easier for her to remember. The partner waits for the patient to contribute the items that she can remember, but provides the missing information when the patient states, “I don’t know Mick’s last name”. Couple 2 showed similar patterns of interaction when completing the First Meeting task together:

In the above dialogue, the couple engage back and forth, with both participants contributing unique information to the recall. The patient interjects several times to clarify their partner’s contributions, but also elaborates on the recall with new information. Later, the partner seeks confirmation from the patient by asking, “Was it a movie?”, which she confirms.

Couple 9’s dialogue during the Mutual Friends task similarly reveals a combination of mirrored repetitions, turn-taking and strategic cueing:

During this exchange, the healthy partner uses categories (“other neighbours”, “your old work people”) to assist in cueing his wife, who responds in each case with unique recall items. There is also evidence of the couple listening and responding to each other’s contributions, through the use of direct questions (“What’s Kelly’s surname?”) and mirrored repetitions (“and Trevor Oakes”).

In contrast to Couples 5, 6 and 7, Couples 1, 2 and 9 showed more symmetric results on the EMQ and PAIR measures. Among these couples, the patients each rated themselves as having a similar frequency of memory failures as their partner’s assessment of their memory (Table 3). In contrast to Couples 5, 6 and 7, they also showed more symmetric results on the PAIR, with the patient in Couple 2 rating their relationship intimacy somewhat lower than their partner, while Couple 1 and Couple 9 were both well matched (Table 2). Therefore, matched ratings of memory function and intimacy may be associated with a ‘stability scaffolding’ style. Interestingly, these cases also showed the relatively rare pattern of patients’ improved recall during collaboration, at least for some tasks, with Cases 2 and 9 recalling more items during collaboration than alone on the CVLT Test 1 (see Table 4), and Case 1 recalling more items during collaboration than alone during the Mutual Friends task (see Table 5), suggesting that their partners were particularly effective at cueing and supporting their recall.

While we could identify instances of these two different scaffolding styles across different couples, we also note that the roles adopted during collaboration and the effects of collaboration varied across tasks. Therefore, it was not straightforward to classify particular couples as always using a particular scaffolding style, or consistently being more or less effective collaborators. Couples did not always adopt the same scaffolding style across all tasks, rather, it appeared to depend on the task at hand, and could also reflect their existing perceptions of relative expertise within the couple. Furthermore, some couples’ collaborative performances did not readily fit into these two proposed scaffolding styles (Cases 3, 4 and 8). Nevertheless, we examined potential differences across measures in these two types of couples.

To assess group level differences between couples identified as adopting ‘survival’ and ‘stability’ scaffolding styles, we conceptually compared patterns of means across a range of measures (see Table 7). Although the numbers of couples in each scaffolding type were small, the differences between these groups are marked. Interestingly, all the couples we identified as ‘stability’ scaffolders were relatively recent post-onset of ABI or epilepsy, while the ‘survival’ group were on average much later post-onset (see Table 7). Perhaps this suggests a trajectory of relationship interaction with the ‘stability’ group still negotiating the recent cognitive changes and reflecting pre-onset interaction styles.

As mentioned above, the healthy partners in ‘survival’ scaffolding couples rated their partners as having greater memory difficulties on the EMQ compared to the ‘stability’ group. This assessment corresponded with actual patient performance on both the CVLT and Mutual Friends tasks, with patients in the ‘survival’ scaffolding group recalling, on average, fewer items during both the individual and collaborative sessions compared to their ‘stability’ scaffolding counterparts (see Table 7). Nevertheless, we did not find evidence of differences between groups during the First Meeting task in Session 2, with ‘survival’ scaffolding patients contributing as many details during collaboration as their ‘stability’ scaffolding counterparts on this task. This suggests that recalling highly personal autobiographical memories (such as a first meeting) with one’s partner might elicit different modes of interaction to more straightforward semantic or verbal learning recall tasks.

To assess the extent to which both patients’ and healthy partners’ assessment of the patient’s memory functioning was associated with actual patient performance, we obtained correlations between patient EMQ scores, partner EMQ scores, and patient recall performance across all couples. There was no relationship between patient EMQ scores and recall performance on any task, (all rs < .361, all ps > .340). That is, patients’ ratings of their everyday memory failures did not reflect their actual memory performance. In contrast, partners’ ratings appeared more consistent with objective performance. There was a significant negative correlation between partner EMQ and patient performance on both the individual Session 1 CVLT-LD, r = –0.672, p = 0.047, and on the patient contribution during Session 2 CVLT-LD, r = –0.739, p = 0.023. Therefore, partners might provide more accurate assessments of memory function than patients themselves. Interestingly, there was a positive correlation between partner EMQ and patient contribution to the First Meeting task in Session 2, where higher scores on the EMQ were associated with more details provided during collaborative recall, r = 0.706, p = 0.034, perhaps reflecting successful scaffolding of performance on this personally relevant task. We did not find significant correlations between partner EMQ measures and patient performance on other tasks.

In summary, patients within ‘survival’ scaffolding couples generally recalled fewer items across the memory tasks, both at the baseline individual level, and during collaborative recall (i.e., they contributed less when working together with their partner). Their generally poorer performance (compared to the ‘stability’ group) was accurately reflected by their partners’ EMQ measures, which rated them as having greater memory difficulties than their own self-ratings. Patient self-ratings on the EMQ did not differ across groups, suggesting that poorer performing ‘survival’ scaffolder patients were not aware of their own memory failures. For the ‘stability’ scaffolding group, measures on the EMQ do not differ between partners and patients, suggesting that each member of the couple has a similar assessment of the patient’s memory functioning. Moreover, ‘survival’ scaffolding couples had highly asymmetric intimacy measures on the PAIR, with healthy partners within ‘survival’ scaffolding couples reporting much lower intimacy assessments than the patient. As with the EMQ results, ‘stability’ scaffolding couples report more symmetric intimacy, with no significant difference between patient and partner PAIR scores among these couples.