Investigating the Role of Imagery Vividness in the Experience of Involuntary Memories of Previously-Imagined Positive Scenarios

Most people experience involuntary memories (IMs) frequently in daily life (Rasmussen et al., 2015), and these are thought to serve important functions, for example directing (or reminding) someone to carry out an action, regulating mood, strengthening a sense of self-identity, and supporting social interactions (e.g. Hashimoto et al., 2022). Most IMs are memories of actual events that have been experienced, but they can also be memories of situations that have only ever been imagined (e.g., Krans et al., 2010). For example, when reading or hearing a news story about a violent assault at a train station someone might find themselves imagining the incident; when they go to the train station themselves the imagined scene may return to mind involuntarily and cause them to be alert to potential dangers. As a more positive example, after reading a sunny weather forecast for the weekend someone might imagine enjoying a barbecue with friends in the garden; if they then encounter one of these friends, this image may then pop into mind and prompt them to invite the friend over. Observational data indicates that IMs can have strong effects on cognition, emotion, and behaviour (see Berntsen, 2021) and there is at least anecdotal data from individuals with depression that this is also the case for IMs of previously-imagined scenes (Blackwell & Holmes, 2010, 2017). Investigating what might increase or decrease the likelihood of an imagined scenario popping back into mind later on may therefore help not only understand a potentially important functional phenomenon in everyday life, but also inform imagery-based clinical interventions for disorders where increasing the frequency of positive imagery may be particularly useful, such as depression.

While most studies investigating IMs have studied those of actually-experienced events, a few have investigated IMs of previously generated mental images. For example, some have asked participants to listen to descriptions of traumatic events such as car accidents and imagine these happening to themselves, before then recording involuntary memories of the imagined scenes in a diary (Krans et al., 2010; Mooren et al., 2019). While the central foci of these studies were not the use of imagery to induce IMs, they did show that mental imagery generation could lead to IMs of the imagined scenes. Further, they investigated some features that may impact involuntary retrieval of the mental image. For example, engaging in a concurrent distracting task while imagining the scenes led to fewer subsequent IMs (Krans et al., 2010), but imagining the scenes from an “observer” (third-person) perspective compared to a “field” (first-person) perspective did not affect the number of IMs (Mooren et al., 2019). Blackwell et al. (2020) investigated involuntary memories of imagined positive events, using an imagery cognitive bias modification (imagery CBM) paradigm, in which participants listened to brief descriptions of everyday scenarios with positive endings and were asked to imagine the scenarios happening to themselves as they unfolded. Participants were randomized to imagine the scenarios either in an emotionally involved or emotionally detached manner, allowing experimental investigation of whether experiencing more intense emotion while imagining the scenarios would lead to greater number of involuntary memories. Contrary to their hypotheses, there were no effects of the experimental manipulation on the number of involuntary memories, either as recorded over the subsequent three days in a diary, or in a lab-based involuntary memory provocation task. However, in exploratory analyses it was found that participants who generated more vivid mental imagery while listening to the scenarios reported a greater number of involuntary memories in the diary (Blackwell et al., 2020).

The exploratory results of Blackwell et al. (2020) raise the possibility that imagining something more vividly might increase the likelihood of that event or scenario later popping back into mind. This could reflect generation of a more vivid image leading to better encoding in memory and greater accessibility. For example, if sensory details in an image are more strongly represented in memory, it may be easier for involuntary recall to be triggered via sensory overlap with an external cue (e.g., seeing one’s front door triggering involuntary recall of a scenario about returning home, in which the front door was part of the image generated). Further, the image recalled as part of the involuntary memory may itself be more vivid, which may be more likely to capture attention and therefore be noticed by an individual when it is retrieved. If the vividness with which someone imagines a positive event happening to them does affect the likelihood of the image later returning involuntarily, then this would have some interesting implications. First, it would open up possibilities for the experimental study of involuntary positive memories, their functions and impact, advancing our theoretical understanding of such phenomena. Despite the assumptions made about the functions of involuntary imagery in daily life, there is little direct evidence for this (Blackwell et al., 2020). An experimental manipulation that could induce a greater number of involuntary memories of a particular valence or content could be used to directly test their impact. Second, if imagining a positive event particularly vividly would make the image more likely to return involuntarily later on, this could have clinical applications. For example, depression is characterized by deficits in positive mental imagery, and this has been suggested to contribute to maintenance of symptoms such as lack of motivation and anhedonia (Holmes et al., 2016; Ji et al., 2019). If deliberate generation of vivid positive mental imagery could lead to later experience of this imagery spontaneously in the context of involuntary memories, this could have beneficial effects, for example in line with the effects on cognition, emotion, and behaviour as reported in the case study by Blackwell and Holmes (2017). In this case study, a depressed participant from a clinical trial of imagery CBM reported that experiencing IMs of the training scenarios (e.g., a scenario about enjoying going for a walk) led to positive expectations about a course of action (e.g., that going for a walk would be enjoyable), positive affect (e.g., anticipatory enjoyment), and behavioural engagement (e.g., going out for a walk). However, the association between vividness and number of IMs found by Blackwell et al. (2020) do not allow causal conclusions about the role of imagery vividness in the later experience of involuntary memories to be drawn; to do so an experimental manipulation of imagery vividness is needed.

In the present study, our main aim was to follow on from the exploratory results of Blackwell et al. (2020) by investigating a method hypothesised to facilitate generation of more vivid imagery, which would then allow direct testing of the hypothesis that imagining a positive scenario more vividly would increase the likelihood that the generated image would later return as an involuntary memory. We therefore carried out a lab-based study in which we used the same imagery CBM paradigm as Blackwell et al. (2020), but tested an experimental manipulation of imagery vividness. To this end, participants were randomized into one of two conditions. As in Blackwell et al. (2020), participants in both conditions listened to brief descriptions of everyday events with positive endings and were asked to imagine them as they unfolded. However, in the present study participants in one condition heard extended versions of the scenarios that had been adapted in a way that we hypothesised would facilitate generation of more vivid imagery (“enhanced condition”). Specifically, the descriptions of the scenarios included more sensory detail, such as emotional and physiological reactions, and detail about both the participant’s and others’ behaviours. The idea that this would lead to more vivid imagery built on work by Peter Lang on the incorporation of further sensory information such as physiological responses into imagery scripts (Lang et al., 1980; Hackmann et al., 2011; see Ji et al., 2016). Participants in the other condition (“standard condition”) heard the “standard” versions of the scenarios as used by Blackwell et al. (2020). To investigate whether this manipulation influenced imagery vividness, participants were asked to rate the vividness of their imagery after each scenario. To verify the basic effects of the imagery CBM (as expected based on previous studies) measures of state positive mood and cognitive bias were collected before after the set of scenarios. During the three days following the lab-based session, participants completed a diary in which they recorded any involuntary memories of the imagined scenarios. Participants attended a second lab-based session three days after the first, at which they completed an involuntary memory provocation task (IMPT; adapted from Berntsen et al., 2013 and Blackwell et al., 2020), designed to evoke involuntary memories of the imagined scenarios. In addition to recording the number of involuntary memories experienced, participants completed measures of state positive mood and negative cognitive bias before and after the IMPT, thus allowing assessment of the effect of IMs on mood and cognition.

To summarise, in the current study we aimed to investigate whether we could influence the vividness of imagine positive scenarios via manipulation of the content provided and whether this in turn would have a causal influence on the later re-experiencing of these positive images in the form of involuntary memories. We also investigated whether we could test effects of the involuntary memories on mood and cognition. Our main hypotheses concerned the effect of the experimental manipulation on imagery vividness and the number of involuntary memories: We predicted that participants in the enhanced condition would rate the vividness of the mental images generated during the imagery CBM as greater than participants in the standard condition. Further, we predicted that participants in the enhanced condition would report more involuntary memories of the imagined scenarios in the three-day diary than participants in the standard condition. We also predicted that the relationship between the experimental condition (enhanced vs. standard) and the number of involuntary memories in the diary would be mediated by vividness of mental images generated during the imagery CBM paradigm.

As secondary hypotheses, we predicted that participants in the enhanced condition would report experiencing more involuntary memories during the IMPT than participants in the standard condition. As we expected that the experience of these positive involuntary memories would be associated with an increase in positive mood and reduced negative interpretation bias (similar to the effect of imagining the scenarios during the imagery CBM), we hypothesized that experiencing a greater number of these involuntary memories during the IMPT would amplify this effect. We therefore predicted that participants in the enhanced condition would show greater increase in positive state mood and reduction in negative cognitive bias, measured before and after the IMPT, than participants in the standard condition.

Eighty students (23 male, 57 female, mean age = 23.32, SD = 3.34, range = 18–34) at Ruhr University Bochum participated in this study. The study was advertised on the psychology faculty website of the university and social media. The inclusion criteria were: being a student aged between 18 and 35, fluent in German, and able to attend both test sessions. A pragmatic sample size of N = 80 complete datasets (i.e. attending both testing sessions and completing the diary; 40 per condition) was planned. This sample size would provide 80% power to detect between-subject effect sizes of Cohen’s d = 0.65 (i.e. medium/large) at level of p <.05, which seemed a plausible estimate given the data on vividness and number of involuntary memories in Blackwell et al. (2020; see Supplementary Materials for further details). At the end of session 2, participants received course credit or €20 for their participation.

The imagery CBM, scrambled sentences test (SST), and involuntary memory provocation task (IMPT) were implemented in Java. Questionnaire measures were completed via an html form in the lab, and the online diary was presented and completed using the Qualtrics ^XM platform.

Imagery Cognitive Bias Modification. The imagery Cognitive Bias Modification (imagery CBM) paradigm was adapted from Blackwell and colleagues (2020). This task has previously been used in experimental and clinical research (e.g. Blackwell et al., 2015). Each trial of the imagery CBM was an audio recording of scenario to which participants were asked to listen via headphones, with their eyes closed, and imagine the scenario happening to themselves as if actively involved and seeing the situation through their own eyes. Before each scenario, there were instructions asking participants to close their eyes and imagine displayed for 1.5 s. All scenarios began ambiguously and had a positive resolution. After each scenario, there was a 2 s pause followed by a beep, which signalled the participants to open their eyes and rate the vividness of their mental image on a 7-point scale from 1 (not at all vivid) to 7 (extremely vivid). While most previous studies used a 5-point scale, including Blackwell et al. (2020), in this study we expanded the scale to 7-points with the aim of achieving more sensitivity to detect the effects of the experimental manipulation on vividness.

The imagery CBM comprised 40 scenarios in total, arranged into 5 blocks of 8 scenarios. Participants could take a short self-paced break between blocks. Half the scenarios had been recorded in a male voice and the other half in a female voice. Prior to the imagery CBM the researcher gave oral instructions and participants practiced mental imagery generation.

Standard Condition. The 40 training scenarios in the standard condition were taken from those used by Blackwell and colleagues (2020), with a few minor alterations to the choice of words (see Supplementary Appendix for the description of standard scenarios). The audio recordings of the scenarios were from 8 to 15 s in length. Here is an example scenario: “You are in the canteen and have lunch alone. You sit down with a group you don’t know well. As they greet you, you realise that they are very pleased that you have joined them.” (positive resolution in italics).

Enhanced Condition. The 40 training scenarios in the enhanced condition were based on the scenarios presented in the standard condition, with additional details included to facilitate generation of more vivid mental images (Lang et al., 1980; Hackmann et al., 2011). Additional details of sensory information (e.g. the visual scene), and what the participant was doing and feeling were added to the start of the scenario (which provides the context and setting). Additional details about the emotions experienced by the participant, bodily sensations, and both the participant’s and others’ behaviour were added to the positive resolution (to reinforce the positive emotional response; see the Supplementary Appendix for the description of scenarios). The addition of these details extended the length of each scenario, meaning that on average the audio recordings in the enhanced condition were twice as long as those in the standard condition. For instance, the ‘canteen’ scenario in this condition was: “You go to the canteen to eat lunch. You look around and notice that no one from your fellow students is there yet. You prepare to eat alone. You sit down at a table where a group of people is already sitting. You don’t know anyone in the group and get a little nervous. As you sit down and they greet you, you realise that they are very pleased that you have joined them. You are relieved and feel happy.” (positive resolution in italics).

Involuntary Memory Diary. Participants were asked to record involuntary memories of the training scenarios in an online diary between session 1 and session 2. Participants received oral instructions for completing the diary from the researcher at the end of session 1. They were requested to report only spontaneous memories and not to report memories resulting from thinking about the scenarios deliberately. Participants were asked to record involuntary memories once per day via a Qualtrics link emailed to them each evening (at about 7:30 p.m.). The link took them to the online diary, where they could record any involuntary memories of the imagery CBM scenarios they had experienced since completing the last diary entry. For each involuntary memory they were asked to specify the time of day they had experienced the memory (morning, afternoon, evening, night), provide a brief summary of the memory (just a few words), and then rate the emotional valence of the memory on an 11-point scale, ranging from − 5 (very negative) to 5 (very positive). In addition, there was a button to click when no involuntary memories had been experienced since the previous diary entry. The first diary link was emailed in the evening following the first lab-based session, and the link was again emailed the subsequent two evenings. If a participant did not complete the diary, they were sent a reminder the following morning and were asked to complete it then; if they had completed it, they were sent a thank you email instead. A final diary entry was completed at the beginning of session 2 in the laboratory (covering the time frame since the diary in the previous evening to the start of the second testing session). To help participants to keep the track of their memories, they were also given a paper pocket diary in which they could briefly note their memories throughout the day. However, this was only for participants’ own use and not collected or enquired about. At the end of session 2, the researcher checked the online diaries with participants. The researcher enquired whether their involuntary memories were from the imagery CBM scenarios, and which scenario each one related to, using the brief description provided in the online diary for guidance. Only the memories of imagery CBM scenarios were included in analyses.

While the previous study (Blackwell et al., 2020) had used a paper-based diary, in line with previous research using the trauma film paradigm (e.g. Holmes et al., 2009), we moved to an online diary in this study to facilitate tracking of data completion. Following research suggesting that different schedules for (electronic) involuntary memory diaries produce similar results (Rattel et al., 2019), we chose a once-per-day schedule for convenience.

Involuntary Memory Provocation Task. The Involuntary Memory Provocation Task (IMPT), adapted from Berntsen et al. (2013) and Blackwell et al. (2020), was used as a means to trigger involuntary memories of the scenarios imagined during the imagery CBM in a controlled laboratory setting and assess their impact on emotion and cognition (see below). The IMPT included 80 trials. Each trial started with a sound clip comprising the initial few words from the imagery CBM scenarios (identical in both experimental conditions), played via either the left or right headphone. After a further 1.5 s, a yellow star was displayed on the left or right side of the screen on a black background. Participants were prompted 500 ms after the star’s appearance to indicate whether the sound clip and the star symbol were on the same side or opposite side by pressing “1” or “2” button on the keyboard, respectively. After the participant gave a response, the next trial started.

To facilitate the aim of using the IMPT to assess the impact of involuntary memories of the imagery CBM scenarios, the task was adapted from the version used by Blackwell et al. (2020) in a number of ways. First, in the current study participants were not asked to report involuntary memories during the IMPT. As reporting involuntary memories increases the duration of the task, this was potentially seen as confounding assessment of the effect of these involuntary memories on mood over the course of the task (as any positive effects of the involuntary memories on mood may be cancelled out by the deleterious effect on mood of the task’s longer duration). Instead, after finishing the IMPT participants retrospectively rated the extent to which they had experienced involuntary memories of the imagery CBM scenarios during the IMPT on a 9-point scale ranging from 1 (not at all) to 9 (all the time). Second, in the current study all the sound clips used were taken from the imagery CBM scenarios, rather than also including other sounds (as in Blackwell et al., 2020), as we aimed to examine the impact of involuntary memories of imagery CBM scenarios specifically (rather than any involuntary memories per se).

Scrambled Sentences Test. The scrambled sentences test (Rude et al., 2002) was used to assess negative interpretation bias, considered to be a cognitive vulnerability to depression. In this study we aimed to use the SST to assess the impact of involuntary memories of the imagery CBM during the IMPT on this cognitive vulnerability measure. Given that imagining the training scenarios during the imagery CBM leads to reductions in negative interpretation bias (e.g., Holmes et al., 2009), we expected that experiencing positive involuntary memories of these training scenarios during the IMPT could also lead to a transient reduction in negative interpretation bias, in this case assessed via the SST. Participants completed the SST before and after the IMPT, and there were two sets of 20 scrambled sentences used. Each sentence had 6 words, and participants were requested to construct a grammatically correct sentence using 5 of these words (e.g. ‘‘winner born I am loser a’’) which could be resolved into a positive (‘‘I am a born winner’’) or negative (‘‘I am a born loser’’) sentence. Participants were asked to construct the first sentence that came to mind and to form a statement, not a question. To reduce participants’ deliberate control over their response, they completed the SST under conditions of cognitive load (keeping a multi-digit number in mind; see next section for details). A “negativity” score was calculated by dividing the number of sentences completed with negative valence by the total sentences completed correctly.

A computerized version of the SST was administered. In general, the computerised format for the SST has been recommended over pen-and-paper formats as this format tends to achieve higher internal consistency (Würtz et al., 2022). The SST used in the current study had been designed to reduce potential practice effects with repeated assessment (and which had been piloted as part of a previous study: Blackwell et al., in press https://​osf.​io/​c752u/​). To this end, the difficulty of the task was tailored to the individual participant and adaptive over the course of the SST, with the aim that the cognitive resources available to participants to reflect on their responses would not increase with practice on the task, but rather than they would continue to provide relatively spontaneous responses. Two aspects of the SST difficulty were continuously varied at participant level: the cognitive load, and the time given to unscramble the sentences.

For cognitive load, a new multi-digit number was shown before every two trials, presented for 5 s. Participants were instructed to remember the number and keep it in mind while unscrambling the sentences. After two trials, participants were asked to enter the number, after which they were presented with a new number for the next two trials. The first number displayed before the first trial included 4 digits. Each time a participant remembered the number correctly twice consecutively, the following number had one more digit (e.g. increasing from 4 to 5 digits). Each time the participant gave two consecutive incorrect responses, the following number had one fewer digit (e.g. decreasing from 4 to 3).

The time available to unscramble a sentence was determined by the duration of their visibility on the screen. The words for the sentence were initially presented in black (surrounded by a black box) on a white background. Over a set time-period for each sentence the colour of the letters faded to white, after which they were no longer visible. At this point only the black surrounding boxes for each word were visible, meaning that the participant had to rely on their memory or guess to provide a response. The first sentence had a time limit of 10 s. Each time a participant unscrambled a sentence correctly and faster than the current time limit, the time taken to unscramble this sentence became the time limit for the subsequent sentence, meaning that the sentence faded out more quickly on the subsequent trial. Each time a participant formed an incorrect sentence, the time limit for the subsequent trial increased by 33%.

Participants completed two sets of the SST (“A” and “B”) in a counterbalanced order (see Supplementary Materials for details). In our sample, the split-half reliability for set A and set B before and after the involuntary memory task were as follows: pre-IMPT version A, α [95% CIs] = 0.71 [0.50, 0.88]; pre-IMPT version B, 0.61 [0.26, 0.87]; post-IMPT task version A, 0.71 [0.47, 0.89]; post-IMPT task version B,0.77 [0.55, 0.92].

On arrival, participants were provided with information about the study. After providing written informed consent, they answered some demographic questions. Participants were randomized to conditions (standard vs. enhanced) at this point (with randomization stratified by gender, testing researcher, and counterbalance order; see Supplementary Materials for detail). Next, they completed the SUIS, PIT, depression and anxiety subscales of DASS-21, past subscale of IAMI, positive subscale of PANAS-X, and AST (version “A” or “B”) in this order. Participants then completed the standard or enhanced imagery CBM, followed by the PANAS-X and then the manipulation check questions. The last task in the first session was the other version of the AST (version “B” or “A”). At the end of session 1, they received the instructions as to how they should record their involuntary memories in the online diary over the subsequent three days.

Three days after the first session, participants returned to the laboratory for session 2. On arrival, they completed the last diary entry. They then completed the PANAS-X, SST (version “A” or “B”), IMPT, and PANAS-X in this order, followed by the question about the experience of involuntary memories of the training scenarios during the IMPT. After that, participants completed the other set of the SST (version “B” or “A”). At the end of session 2, participants were debriefed and received course credit or cash payment for their participation.

To comply with Transparency and Openness Promotion (TOP; Nosek et al., 2015) and reproducibility guidelines, we report sample size calculation, tasks, and questionnaires used in the study. The study materials, data, and data analysis scripts are available on the Open Science Framework (OSF) via this link (https://​osf.​io/​jhs37/​). The study was preregistered on the OSF prior to starting data collection (https://​osf.​io/​57a89). The study protocol, including design, hypotheses, tasks and questionnaires, and data analysis plan was uploaded on the OSF before starting data collection (https://​osf.​io/​jhs37/​).

The present study investigated the role of imagery vividness in the experience of involuntary memories of previously-imagined positive scenarios. Participants completed an imagery cognitive bias modification (imagery CBM) task in which they listened to audio recordings of everyday scenarios with positive endings and imagined them as they unfolded, with half of the sample hearing ‘enhanced’ versions of the scenarios hypothesised to facilitate generation of more vivid imagery. Contrary to our prediction, there were no differences between participants hearing the enhanced scenarios (‘enhanced condition’) and participants hearing the standard versions of the scenarios (‘standard condition’) in how vividly they imagined the scenarios. Further, there was no difference between the two conditions in the number of involuntary memories of the scenarios experienced, either in daily life over the next three days or in a lab-based involuntary memory provocation task (IMPT). However, the more vividly participants imagined the scenarios during the imagery CBM, the greater the number of IMs of the scenarios they recorded in the diary. While the lack of effect of the experimental manipulation on imagery vividness means that the current study could not directly test the causal role of vividness in later experience of involuntary memories of positive images, the findings shed further light on factors associated with the frequency and characteristics of these memories.

In the current study we aimed to facilitate generation of more vivid imagery during the imagery CBM by adding further information to the descriptions of the scenarios participants were asked to imagine. Specifically, we added more detailed sensory information about the setting, details of actions carried out by the participant and others, and the participant’s emotional and physiological reactions. However, this did not lead to generation of more vivid mental images. A possible explanation for this is that within the constraints of the imagery CBM task, in which the scenarios and time for imagining are relatively brief (up to ~ 15 s for the standard scenarios or up to ~ 30 s for the enhanced scenarios), there may be insufficient time to benefit from additional sensory information. In contrast, in the previous work on which we based this manipulation, the time for imagery generation was much longer or unconstrained (Lang et al., 1980; Hackmann et al., 2011). Further, the descriptions in the “standard” scenarios were devised to be easily imaginable, and generally took place in familiar contexts for which participants would have highly accessible sensory information available in memory. Providing additional information may therefore have added little above what could already be readily retrieved from memory. Within the constraints of the imagery CBM, the vividness of imagery generated may rather be more constrained by the participants’ general ability to generate positive imagery, as indicated by the correlation between vividness ratings made during the imagery CBM and those for positive items on the prospective imagery test (PIT; see Supplementary Material). It is possible that increases in imagery vividness during the imagery CBM could be achieved via other means, e.g., extended training and prior practice in imagery generation, or more time for imagery elaboration after hearing the scenario descriptions. However, it may be that experimental manipulations aiming to reduce the vividness of the mental imagery generated would be more successful. For example, participants could be asked to do a concurrent task engaging visuo-spatial working memory during mental imagery generation (Lilley et al., 2009) or afterwards (Holmes et al., 2009). However, regardless of the effect of the manipulation on vividness, our results demonstrate that adding further sensory details to the scenarios to be imagined did not lead to a greater number of involuntary memories.

Given that there was no difference in vividness ratings made during the imagery CBM between the two experimental conditions it is perhaps not surprising that there was also no difference between the conditions in the number of IMs recorded in the diary or IMPT. Interestingly, and replicating the exploratory results of Blackwell et al. (2020), we found that participants with higher mean vividness ratings for the scenarios during the imagery CBM recorded more IMs in the diary. Potentially this implies that if we had successfully managed to manipulate imagery vividness, we might have seen between-group differences in the number of IMs, but this is yet to be demonstrated and the relationship between imagery vividness and IM frequency remains correlational. Notably there were methodological differences between the current study and that of Blackwell et al. (2020), in that we used a 7-point, rather than a 5-point, vividness scale, and an electronic, rather than paper-based, involuntary memory diary. Our replication of the previous results despite these differences can increase our confidence in the robustness of the correlational relationship between imagery vividness and frequency of later IMs.

One intriguing finding in the current study was that those participants who reported a greater increase in positive affect during the imagery CBM recorded more IMs in the diary. This fits with the results of previous research using negative (Clark et al., 2015) and positive (Clark et al., 2013) film stimuli to induce involuntary memories, in which mood change while watching the films predicted the number of subsequent IMs. However, this result stands in contrast to the findings of Blackwell et al. (2020), who found no such relationship. Blackwell et al. (2020) speculated that this could have been due to including several tasks (a voluntary memory test and an involuntary memory provocation task) in the testing session after the imagery CBM. They hypothesised that these tasks could have interfered with encoding of the scenarios into memory and any potential effect of emotion on the number of IMs. One implication of our current results is that an experimental manipulation that increased the emotion experienced by participants while imagining the scenarios (as opposed to imagery vividness) could indeed provide a means to increase the number of IMs experienced. Instructions to focus on simulating the experience of the emotions described in the scenarios (as opposed to focussing primarily on creating a detailed visual scene) could provide one simple route to achieving this.

In the current study we aimed to detect the effects of involuntary memories of previously-generated positive imagery on mood and cognition by provoking IMs via audio cues in the lab-based IMPT. However, our results provide no evidence for such effects, and it seems plausible that exposure to the audio cues and the IMs experienced had little impact. In fact, most participants reported they did not experience IMs of the scenarios during the IMPT; a lack of effect on mood and cognition is therefore unsurprising. Further, the scenarios related to relatively everyday mildly positive experiences; it might be that in a sample of mostly healthy university students, as in the current study, more emotionally intense scenarios would be required for IMs of them to have an impact. It is also possible that the IMs had an impact on participants’ mood but that this was transient and would have needed assessment directly after the experience of an IM to be detectable. Future studies may therefore benefit by asking participants to record mood more proximally to the occurrence of IMs, e.g., immediately after experiencing an IM or more regularly throughout the IMPT. It may also be easier to investigate the impact of IMs of positive images amongst participants for whom their occurrence would be more noticeable, e.g., people with depression or low mood.

The results of the current study have potential implications for both experimental and clinical studies. They add to those of Blackwell et al. (2020) by providing a further demonstration that the imagery CBM paradigm can be used as a tool to induce positive involuntary imagery in people’s daily lives. From an experimental perspective, this opens possibilities for investigating both modulators and effects of involuntary imagery, including testing of causal hypotheses via experimental manipulation of the content or frequency. From a clinical perspective, the results add to anecdotal reports from participants with depression in clinical studies of imagery CBM that IMs of the scenarios can occur, and support investigating this phenomenon as a potential treatment mechanism. The results suggest that it may be worthwhile investigating not only increasing vividness but also increasing emotional engagement in imagery as a means to increase the frequency of IMs experienced.

The results of the current study must be interpreted within the context of a number of limitations. First, as our first hypothesis was not confirmed, in that we did not find the predicted effect of our experimental manipulation on vividness, we were not able to directly test the causal role of vividness in the occurrence of IMs. Second, the number of IMs recorded was small, and it is not clear whether this reflects an actual absence of IMs or participants not recording IMs. Third, while the used of an unselected (predominantly healthy) sample should allow us to investigate mechanisms involved in everyday positive IMs, it makes it harder to detect their potential impact on mood and does not necessarily allow extrapolation to clinical samples. Fourth, because of the novelty of the combination of tasks (e.g., imagery CBM, the diary, and IMPT), they have by necessity not been validated in this form and this must be taken into account in interpreting the findings. Further, internal consistency was only moderate for some measures administered, which may have reduced their sensitivity. Fifth, although our consideration of the data from a previous similar study (Blackwell et al., 2020) suggested that our sample size provided sufficient statistical power, it is also possible that the effect sizes under investigation were smaller than expected and a larger sample would be needed to detect them. Additionally, our study was not specifically powered to analyse the relationship between imagery vividness and number of IMs, although we note that we find a similar effect size (odds ratio) for this analysis to that found by Blackwell et al. (2020). Sixth and finally, the researchers administering the study procedures were not blind to participant allocation once a participant had completed the imagery CBM as the enhanced condition took longer to complete, and this could provide a source of bias in the results (albeit in the absence of between-group differences this seems unlikely).

The current study sheds some further light on factors that are associated with the likelihood that imagined positive events are later re-experienced in the form of involuntary memory. Although we cannot draw causal conclusions about vividness from the findings, the results suggest that generating more vivid mental imagery and experiencing a greater emotional response to this imagery is associated with greater likelihood of subsequent involuntary recall, and thus both imagery vividness and emotional response to imagery should be targets for future work aiming to establish causal relationships. Further investigation of involuntary recall of previously-generated images could help further our understanding of a common and potentially important phenomenon in everyday life, as well as help develop and improve imagery-based interventions for disorders characterised by deficits in positive imagery such as depression.