Introduction
Episodic future thinking, or episodic simulation, refers to the ability to project oneself into the future within specific scenarios, often recruiting mental imagery (Schacter et al., Reference Schacter, Benoit and Szpunar2017). Functional accounts propose that this ability is crucial for a range of skills, including planning, problem solving and emotional regulation (Szpunar et al., Reference Szpunar, Spreng and Schacter2014). While future thinking is thought to emerge early on in childhood (Steinberg et al., Reference Steinberg, Graham, O’Brien, Woolard, Cauffman and Banich2009), adolescence is arguably a key stage for its development. This period coincides with the critical life task of negotiating one’s past identity with future goals, in relation to education, career, relationships, and so forth (Johnson et al., Reference Johnson, Blum and Cheng2014; Nurmi, Reference Nurmi1991). Simultaneously, adolescence is also a period of increased emotionality and vulnerability when most mental health problems first emerge (Solmi et al., Reference Solmi, Radua, Olivola, Croce, Soardo, Salazar de Pablo, Il Shin, Kirkbride, Jones, Kim, Kim, Carvalho, Seeman, Correll and Fusar-Poli2021). Taken together, there is a need to better understand the role of future thinking in adolescence and its link to psychopathology.
Clinical and fundamental research on involuntary prospective imagery
Recent neuroscientific developments highlight the role of prospection in the simulation of threats (Bulley et al., Reference Bulley, Henry and Suddendorf2017). A particular form of prospection, intrusive prospective imagery, is hypothesised to play an important role in the aetiology, maintenance and recovery of emotional disorders (Brewin et al., Reference Brewin, Gregory, Lipton and Burgess2010). Such mental images, also known as ‘flashforwards’ (Engelhard et al., Reference Engelhard, Van den Hout, Janssen and van der Beek2010; Hales et al., Reference Hales, Deeprose, Goodwin and Holmes2011), are involuntarily generated, perception-like, and emotionally charged. Flashforwards are common in anxiety disorders, where images frequently depict negative feared catastrophes. Examples include picturing oneself dying of cancer in health anxiety (e.g. Lau-Zhu and Brummer, Reference Lau-Zhu and Brummer2022; Muse et al., Reference Muse, McManus, Hackmann, Williams and Williams2010), being ridiculed by peers at an incoming party in social phobia (Chiu et al., Reference Chiu, Clark and Leigh2022; Frets et al., Reference Frets, Kevenaar and Van Der Heiden2014; Reimer and Moscovitch, Reference Reimer and Moscovitch2015), or harming a loved one in obsessive compulsive disorder (e.g. Lau-Zhu et al., Reference Lau-Zhu, Farrington and Bissessar2022; Rachman and de Silva, Reference Rachman and de Silva1978). These flashforwards are thought to amplify threat appraisals and maladaptive coping (e.g. avoidance), maintaining anxiety problems (Brewin et al., Reference Brewin, Gregory, Lipton and Burgess2010).
Involuntary mental images stand in contrast to those that are voluntarily generated (Lau-Zhu et al., Reference Lau-Zhu, Henson and Holmes2019; Lau-Zhu et al., Reference Lau-Zhu, Henson and Holmes2021; Pearson and Westbrook, Reference Pearson and Westbrook2015) and which have been traditionally neglected in mainstream cognitive research (Finnbogadóttir and Berntsen, Reference Finnbogadóttir and Berntsen2013). However, there has been a recent surge in scientific interest on involuntary forms of future thinking in everyday life, which are often imagery-rich (Barsics et al., Reference Barsics, Van der Linden and D’Argembeau2016; Cole and Kvavilashvili, Reference Cole and Kvavilashvili2019). Spontaneous thoughts in non-clinical populations are more often characterised by positive (rather than negative) future projections (Barsics et al., Reference Barsics, Van der Linden and D’Argembeau2016; Finnbogadóttir and Berntsen, Reference Finnbogadóttir and Berntsen2013). This positive bias may reflect a basic motivational tendency to seek out positive information to maintain a positive self-view, while distancing from threatening materials (Finnbogadóttir and Berntsen, Reference Finnbogadóttir and Berntsen2013). The increased incidence or intensity of spontaneous positive imagery has been associated with higher levels of optimism (Beaty et al., Reference Beaty, Seli and Schacter2019) and lower levels of depressive symptoms (Ji et al., Reference Ji, Holmes, MacLeod and Murphy2019). Therefore, spontaneous future thinking arguably plays important functions including goal pursuits and emotional regulation (Barsics et al., Reference Barsics, Van der Linden and D’Argembeau2016).
While experiencing spontaneous positive images can be beneficial (Barsics et al., Reference Barsics, Van der Linden and D’Argembeau2016), in some cases they can also contribute to distress. Emotional dysregulation (in mood and anxiety disorders) has been associated with increased emotional impact of flashforward images of negative and positive events (Di Simplicio et al., Reference Di Simplicio, Renner, Blackwell, Mitchell, Stratford, Watson, Myers, Nobre, Lau-Zhu and Holmes2016; Di Simplicio et al., Reference Di Simplicio, Lau-Zhu, Meluken, Taylor, Kessing, Vinberg, Holmes and Miskowiak2019; van den Berg et al., Reference van den Berg, Voncken, Hendrickson, Houterman and Keijsers2020). Overly positive images can be found in bipolar mania (Ivins et al., Reference Ivins, Di Simplicio, Close, Goodwin and Holmes2014) and cravings in addictions (Kavanagh et al., Reference Kavanagh, Andrade and May2005).
Emerging research in youth populations
Intrusive emotional imagery research has so far been mostly restricted to adults (Brewin et al., Reference Brewin, Gregory, Lipton and Burgess2010), even though adolescents appear to be more vivid imagers than adults (Gulyás et al., Reference Gulyás, Gombos, Sütöri, Lovas, Ziman and Kovács2022). Studies with youth samples are, however, beginning to accumulate (Burnett Heyes et al., Reference Burnett Heyes, Lau and Holmes2013; Lau-Zhu et al., Reference Lau-Zhu, Farrington and Bissessar2022; Pile et al., Reference Pile, Williamson, Saunders, Holmes and Lau2021; Schwarz et al., Reference Schwarz, Grasmann, Schreiber and Stangier2020). The majority have focused on social anxiety (for a review, see Chapman et al., Reference Chapman, Halldorsson and Creswell2020), revealing that adolescents with social anxiety disorder report intrusive imagery of feared social scenarios, which maintains anxiety and contaminates social interactions (Leigh et al., Reference Leigh, Chiu and Clark2020), although the time orientation of such imagery is not always queried.
In a recent study specifically querying for the future, greater impact of flashforwards of negative future events (e.g. ‘my dog getting ill’ or ‘my parents arguing’) was linked to more symptoms of generalised anxiety and depression in a community sample of adolescents (Pile and Lau, Reference Pile and Lau2020). However, measurement of flashforwards here combined ratings of imagery with their associated responses (i.e. resulting arousal and avoidance), leaving it unclear whether and which aspect of flashforward imagery independently contributed to symptomatology. Flashforwards reported were also restricted to negative events, whereas recent work suggests that responses to both negative and positive future events (i.e. a general propensity to emotional imagery) can be pertinent to youth distress (Deeprose et al., Reference Deeprose, Malik and Holmes2011; O’Donnell et al., Reference O’Donnell, Di Simplicio and Burnett Heyes2020).
Finally, prospective positive imagery has been investigated during daily life (Marciniak et al., Reference Marciniak, Shanahan, Binder, Kalisch and Kleim2023). This study, however, focused on deliberate generation (prompted by an app) rather than naturally occurring, involuntary generation central to the phenomenology of flashforwards.
Developmental differences and psychopathology vulnerabilities
Existing research on involuntary mental images in youth has mainly focused on typically developing individuals (with anxiety problems). Our understanding of this form of cognition in those with developmental differences, which can heighten emotional vulnerabilities and often present to clinical services, remains limited. Here we consider autism spectrum and maltreatment history – two developmental presentations frequently overlooked in this line of research. We put forward three compelling reasons for considering these populations.
First, from an intervention standpoint, both autism and maltreatment are associated with elevated rates of mental health problems including anxiety and depression (Gilbert et al., Reference Gilbert, Widom, Browne, Fergusson, Webb and Janson2009; Lai et al., Reference Lai, Kassee, Besney, Bonato, Hull, Mandy, Szatmari and Ameis2019) but with reduced benefits from gold-standard psychological interventions (Lippard and Nemeroff, Reference Lippard and Nemeroff2020; Weston et al., Reference Weston, Hodgekins and Langdon2016). This clinical challenge suggests that imagery may represent a neglected target for improved interventions in these groups. Informed by existing cognitive accounts, one could predict that involuntary generation of emotional events (of both past and future orientation) is heightened in both autistic and maltreatment-exposed individuals relative to their typically developing peers, albeit through different mechanisms. In autism, reduced executive functioning could impair one’s ability to adaptively inhibit unwanted intrusions (McDonnell et al., Reference McDonnell, Valentino and Diehl2017), leading to heightened flashforwards. In contrast, after maltreatment, excessive reliance on avoidance coping over time turns into a habitual cognitive style for managing threatening material, which paradoxically exacerbates intrusions (Hitchcock et al., Reference Hitchcock, Werner-Seidler, Blackwell and Dalgleish2017), including flashforwards.
Two studies hint at a potential heightened intrusion profile in these populations but have not probed underlying mechanisms. Ozsivadjian et al. (Reference Ozsivadjian, Hollocks, Southcott, Absoud and Holmes2017) found that autistic children reported more involuntary anxious images, but the future was not specifically queried. Steil et al. (Reference Steil, Fischer, Gutermann and Rosner2022) showed that maltreatment-exposed adolescents reported more negative involuntary images, but did not separate past and future events in their analyses.
Second, there is a pervasive assumption that autistic individuals are poor imagers (Dance et al., Reference Dance, Jaquiery, Eagleman, Porteous, Zeman and Simner2021) with limited imagination (Baron-Cohen et al., Reference Baron-Cohen, Wheelwright, Skinner, Martin and Clubley2001), leading many clinicians to question the applicability of imagery-based interventions to improve autism mental health. While studies like that of Ozsivadjian et al. (Reference Ozsivadjian, Hollocks, Southcott, Absoud and Holmes2017) are beginning to challenge these assumptions, additional evidence is required.
Third, autism and maltreatment-related presentations have traditionally been studied separately, but there is a growing recognition of the need to understand their overlaps and distinctions (Dinkler et al., Reference Dinkler, Lundström, Gajwani, Lichtenstein, Gillberg and Minnis2017; Gajwani and Minnis, Reference Gajwani and Minnis2023), especially in relation to cognitive profiles and processes (Lau-Zhu et al., Reference Lau-Zhu, Chan, Gibson, Stark, Wang, Happé, Stacey and Cooper2024). This comparison is crucial because both conditions present with overlapping difficulties often complicating differential diagnosis and in turn the provision of appropriate support (Davidson et al., Reference Davidson, Moran and Minnis2022; Moran, Reference Moran2010).
The present study
We sought to investigate flashforwards in adolescents aged 10–16, enriched with developmental differences, (i) to explore flashforward characteristics without constraining them for the first time, such as to a specific valence (Pile and Lau, Reference Pile and Lau2020) or anxious scenarios only (Ozsivadjian et al., Reference Ozsivadjian, Hollocks, Southcott, Absoud and Holmes2017). We then probed for (ii) differences in flashforwards between autistic adolescents, maltreatment-exposed adolescents, and adolescents without autism or maltreatment (typically developing adolescents for brevity); and (iii) associations between flashforward characteristics and symptoms of anxiety/depression; and (iv) associations between flashforward characteristics and executive functioning/avoidant coping.
As primary hypotheses, we predicted that (i) flashforward characteristics are heightened in autistic and maltreatment-exposed adolescents compared with typically developing adolescents and (ii) flashforward characteristics are associated with mental health (anxiety and depression).
As an exploratory hypothesis, we predicted that (iii) such flashforward characteristics are associated with avoidant coping (especially in the maltreatment-exposed group) and executive dysfunction (especially in the autism group).
Method
Participants
Eighty-seven UK-residing participants (aged 10–16) and their caregivers or parents took part. Adolescents aged below 16 provided online assent. Parents, caregivers and legal guardians and adolescents aged 16 provided online consent. For each adolescent, a small donation of £2 was given to a youth charity either working with the autism community (Parents Talking Asperger’s) or young people who had experienced significant early adversities (SAFE!). Data on different results from this same sample have been presented elsewhere (Lau-Zhu et al., Reference Lau-Zhu, Chan, Gibson, Stark, Wang, Happé, Stacey and Cooper2024).
Recruitment sources included local child and adolescent mental health services, children’s social services, and community advertisement which included charities and university webpages. Adolescents in the autism group (n=30) had a confirmed diagnosis of ASD based on the Diagnostic and Statistical Manual (DSM) 4th edition (American Psychiatric Association, 1994) or 5th edition (American Psychiatric Association; 2013), or the International Classification of Diseases (ICD) 10th edition (WHO; World Health Organization, 2016) or 11th edition (World Health Organization, 2019), by their keyworker (who had access to clinical records) or caregiver (who provided written evidence, if needed), and who also confirmed there was no indication of a history of maltreatment or prior social service contact due to the child’s quality of care. Adolescents in the maltreatment group (n=28) had a documented history of maltreatment (i.e. sexual, physical or emotional abuse, and/or neglect) confirmed by their keyworkers, who also confirmed the absence of a (suspected) ASD diagnosis or first/second-degree relatives with ASD. Adolescents in the typically developing group (n=29) had their caregivers confirm the absence of indicators (as described above) of either ASD or maltreatment history. Adolescents were not eligible if they: (1) did not read English; (2) had an organic brain condition (e.g. brain injury); (3) had experiences of psychotic episodes; (4) had a diagnosis of a learning disability or known IQ<70; (5) were actively suicidal; and/or (6) had sensory impediments that would interfere with completing questionnaires (e.g. difficulties with screen exposure >15 min).
Measures
Adolescents and caregivers completed an online survey each (with self-reported measures) separately between July 2021 and April 2022.
Flashforwards survey
We integrated different approaches to assessing emotional imagery from previous work. Survey instructions and format were adapted for a study conducted during the COVID-19 pandemic in the UK, in consultation with autistic and maltreatment-exposed adolescents, their caregivers, and professionals working with them, to ensure understanding and feasibility while minimising burden.
First, we drew items from a youth version of the Impact of Future Event Scale (following Pile and Lau, Reference Pile and Lau2020), which in turn was adapted from the original adult version (Deeprose and Holmes, Reference Deeprose and Holmes2010). Instruction began as follows: ‘The next questions are about how you think about the future. “Future” means anything that can happen tomorrow, or in the next few days, next few weeks, next few months, next few years, or even when you get much older. For each question, select how often it applies to you. The questions are about your experiences in the past 7 days’. Participants were then presented with four items of the intrusion subscale (e.g. ‘Do mental pictures about the future pop into your mind?’ and ‘Do you think about the future even when you don’t mean to?’), with four response options (‘not at all’, ‘rarely’, ‘sometimes’ and ‘often’). A total frequency score was the sum of all items, with values ranging from 0 to 20. This subscale showed good internal consistency in this study (Cronbach’s α=0.73).
Second, participants were asked to provide a brief (the most important) example of what it is that they had been thinking might happen in the future for them. These responses were reviewed by the first author (A.L.-Z., who was blinded to group membership during coding) to categorise them based on previous research on spontaneous imagery in youth (Ozsivadjian et al., Reference Ozsivadjian, Hollocks, Southcott, Absoud and Holmes2017; Pile and Lau, Reference Pile and Lau2020). Twenty per cent of these were double-coded (by D.G.) with excellent agreement, Cohen’s kappa = 0.96 (Altman, Reference Altman1991).
Third, participants answered four questions (following the Imagery Interview; Di Simplicio et al., Reference Di Simplicio, Renner, Blackwell, Mitchell, Stratford, Watson, Myers, Nobre, Lau-Zhu and Holmes2016; Di Simplicio et al., Reference Di Simplicio, Lau-Zhu, Meluken, Taylor, Kessing, Vinberg, Holmes and Miskowiak2019; Ozsivadjian et al., Reference Ozsivadjian, Hollocks, Southcott, Absoud and Holmes2017) about additional phenomenological qualities of the reported significant image: (i) valence (‘Is this event positive or negative?’; response options were ‘positive’, ‘negative’ and ‘none’); (ii) vividness (‘How clear was this future event in your head when you imagined it?’); (iii) emotionality (‘How emotional did you feel when you imagined it?’); (iv) likelihood (‘How likely did you think this event was to happen?’). Response options (except for valence) used Likert scales of 1–5 anchored from ‘not at all’ to ‘extremely’.
Our assessment of flashforwards differed from prior work in two key ways. First, we did not include the arousal/avoidance subscales (Pile and Lau, Reference Pile and Lau2020) as our main interest was in assessing imagery characteristics rather than its impact (and to reduce questionnaire length and construct overlap with measures of anxiety). Second, we did not constrain the intrusion scale to only negative events as we were interested in adolescents’ general propensity towards involuntary prospective imagery, which is more in line with the original conceptualisation in adults where both positive and negative events are included (Deeprose and Holmes, Reference Deeprose and Holmes2010). Finally, we considered a range of phenomenological characteristics beyond frequency of occurrence, in line with interview studies (Di Simplicio et al., Reference Di Simplicio, Renner, Blackwell, Mitchell, Stratford, Watson, Myers, Nobre, Lau-Zhu and Holmes2016; Di Simplicio et al., Reference Di Simplicio, Lau-Zhu, Meluken, Taylor, Kessing, Vinberg, Holmes and Miskowiak2019).
Questionnaires
For adolescents
Anxiety and depression symptoms were assessed with the Revised Children’s Anxiety and Depression Scale-11 items (RCADS-11; Radez et al., Reference Radez, Waite, Chorpita, Creswell, Orchard, Percy, Spence and Reardon2021). PTSD symptoms were assessed with the Children’s Revised Impact of Event Scale (CRIES; Perrin et al., Reference Perrin, Meiser-Stedman and Smith2005), yielding separate subscales for re-experiencing and avoidance, the latter serving as an index of avoidant coping (Kuyken et al., Reference Kuyken, Howell and Dalgleish2006). A proxy for general cognitive ability (GCA) was performance using the Abbreviated 9-item form of the Raven’s Standard Progressive Matrices Test (RSPMT-9; Bilker et al., Reference Bilker, Hansen, Brensinger, Richard, Gur and Gur2012), which is highly predictive of the original 60-item form (Raven, Reference Raven2000) and recently used in adolescents (e.g. Bone et al., Reference Bone, Lewis, Roiser, Blakemore and Lewis2021; Morin et al., Reference Morin, Afzali, Bourque, Stewart, Séguin, O’Leary-Barrett and Conrod2019). Trait (non-future) mental imagery vividness was assessed with the Plymouth Sensory Imagery Questionnaire, where adolescents rated vividness to imagined items across different sensory modalities (Andrade et al., Reference Andrade, May, Deeprose, Baugh and Ganis2013).
For caregivers
Caregivers completed questionnaires in relation to the child. Anxiety and depression symptoms were assessed with the RCADS-47 (Chorpita et al., Reference Chorpita, Moffitt and Gray2005). Caregiver-rated PTSD symptoms based on DSM-5 criteria (American Psychiatric Association, 2013) were assessed with the Child and Adolescent Trauma Screen (CATS; Sachser et al., Reference Sachser, Berliner, Holt, Jensen, Jungbluth, Risch, Rosner and Goldbeck2017). Autistic traits were assessed with the Social Communication Questionnaire-Current version (SCQ; Rutter et al., Reference Rutter, Bailey and Lord2003) with a cut-off score of ≥15 indicative of probable ASD (Chesnut et al., Reference Chesnut, Wei, Barnard-Brak and Richman2017). Everyday executive functioning was assessed with the Dysexecutive Questionnaire-Children (DEX-C; Emslie et al., Reference Emslie, Wilson, Burden, Nimmo-Smith and Wilson2003). Information on adolescents’ demographics, known diagnoses, and parental/caregiver highest educational level were also collected (the latter as a proxy for socioeconomic status or SES; Liberatos et al., Reference Liberatos, Link and Kelsey1988).
Statistical analyses
Histograms were inspected to assess for normality. Homogeneity of variance was assessed using the Levene’s statistic. For continuous variables, overall differences across the three groups were assessed with one-way ANOVAs and follow-up independent-sample t-tests as appropriate. For categorical variables, group differences were assessed with chi-square tests. A two-tailed alpha level of .05 was used. Associations between flashforward characteristics and other variables (e.g. mental health) used Pearson’s correlation tests. Principal component analyses (varimax rotation and eigen value >1) were used to extract a common ‘depression’ and ‘anxiety’ components combining the adolescent and caregiver versions of the RCADS to generate a single score for each, therefore minimising the number of correlations performed. Sizes of correlations were compared with Fisher’s tests. Hierarchical multiple regressions were used to predict the influence of flashforward characteristics on anxiety and depression, after controlling for baseline variables that correlated with predictors and/or outcomes (i.e. age, sex, and trait imagery vividness). All statistical analyses were conducted in SPSS version 27 (IBM, 2020).
Results
Basic demographics and clinical profiles
There were no statistical differences between the three groups in terms of age, sex, SES and trait imagery vividness (across sensory modalities), although the proportion of adolescents described as of White ethnicity was highest in the autism group. Both the autism and maltreatment groups scored lowest on a caregiver-reported measure of executive functioning, while only the maltreatment group scored lowest on a measure of GCA. The number of autistic traits was highest in the autism group, followed by the maltreatment group and then the typically developing group. The number of past traumatic events was highest in the maltreatment group, followed by the autism group and then the typical group. Symptoms of emotional disorders were also highest in the autism and maltreatment groups. See Table 1.
Table 1. Background variables including demographics and clinical measures by group
TD; typical development; RSPMT-9, Raven’s Standard Progressive Matrices Test 9-Items Short Form; DEX-C, Dysexecutive Questionnaire Child Version; Psi-Q, Plymouth Sensory Imagery Questionnaire; SCQ, Social Communication Questionnaire Current Version. a,b,cDifferent letters are used to denote significant pairwise differences; doverall group comparisons (i.e. across the three groups); eone participant in the autism group (female at birth) identified as transgender; fdiagnoses of neurodevelopmental disorders included ADHD, Tourette’s syndrome, dyspraxia, dyscalculia, dyslexia, and sensory processing disorder; gdiagnoses of emotional disorders included post-traumatic stress disorder, obsessive compulsive disorder, body dysmorphic disorder, depressive disorder, generalised anxiety disorder; hmissing data were found for SES (maltreatment=4; TD=1); RSPMT-9 (autism=3, maltreatment=1, TD=1); Psy-Q (autism=3, maltreatment=1, TD=2); SCQ (maltreatment=2); RCADS-11 (autism=1; TD=1); DEX-C (autism: n=1, maltreatment: n=2); CRIES (autism=2; TD=1).
To explore for potential covariates (see Table 2), across the whole sample, no significant correlations were found between clarity (vividness), emotionality and likelihood ratings of flashforwards and age, sex, ethnicity, SES or GCA. However, frequency of flashforwards significantly and positively correlated with age and sex, indicating that older and female adolescents reported higher occurrence of flashforwards. Trait imagery vividness was also positively correlated with clarity and likelihood ratings for flashfowards.
Table 2. Correlation matrix for key baseline variables, flashforward characteristics, and mental health, for the full sample
SES, socioeconomic status, indexed by parental education (university vs non-university); GCA, general cognitive ability indexed by Raven’s Standard Progressive Matrices Test 9-Items Short Form; FF, flashforward; anxiety and depression scores were derived by combining self-report (RCADS-11) and caregiver-reports (RCADS-47) as a factor score; n per correlation varied from 82 to 87 due to missing data; *p<.05, **p<.01, ***p<.001.
Content of flashforwards
For valence, over half (56.3%) of flashforwards were endorsed as of ‘positive’ valence (20.7% as ‘negative’ and 23% reported as ‘none’) across groups. Thematically, most flashforwards across groups were about ‘career, education and/or learning’ (62.3%; e.g. ‘I may get a job as a scientist’; ‘moving to secondary school’; ‘I will be a dancer’) followed by ‘family, friends, and/or relationships’ (20.8%; e.g. ‘imagining my future wife’; ‘getting another sibling’; ‘living with my mum’). A small proportion was on other themes, including distress (6.5%; e.g. ‘anxious about what is going to happen in school’), leisure (6.5%; e.g. ‘getting really good at scootering’) and society (3.9%; e.g. ‘when cars are autopiloted’). The distributions of image types in terms of valence or themes did not significantly differ between groups, χ2‘s<11.78, p>.162.
Finally, only a small proportion was thematically linked to concerns typical of DSM-5 anxiety disorders (26.6%; ‘having a bad reaction to COVID-19 and ending up in hospital’), DSM-5 traumatic events (7.6%; ‘dying traumatically’), or depicting interpersonal scenarios (36.7%; ‘people cancelling plans’).
Phenomenological characteristics of flashforwards
Contrary to hypothesis, groups did not significantly differ on any flashforward characteristic considered (Table 3), including image clarity, F 2,84=2.22, p=.115, ηp 2=0.05, emotionality, F 2,84=2.06, p=.134, ηp 2=0.05, likelihood, F 2,84=1.17, p=.317, ηp 2=0.03, and frequency, F 2,82=1.12, p=.317, ηp 2=0.03. We repeated these analyses with positive images only (as these were the majority of images reported), but the same pattern of results remained. Excluding autistic participants with low SCQ scores (<15; Chesnut et al., Reference Chesnut, Wei, Barnard-Brak and Richman2017) or maltreatment-exposed participants with high SCQ scores (15 and above; Chesnut et al., Reference Chesnut, Wei, Barnard-Brak and Richman2017) did not change the pattern of findings.
Table 3. Means and standard deviations for flashforward characteristics by group
TD, typically developing; FF, flashforward.
Associations with mental health
See Table 2 for correlational analyses. Partially consistent with hypotheses, flashforward emotionality was positively correlated with both anxiety and depression. Flashforward frequency was also positively correlated with both anxiety and depression. However, correlations between flashforward clarity/likelihood and anxiety/depression were not significant. When considering only adolescents who reported positive mental images, all the above correlations were no longer significant, except for a negative correlation between flashforward likelihood and depression, r(49)=–0.35, p=.013.
Hierarchical multiple regressions (see Table 4) were performed for anxiety and depression factors separately. In the first step, age, sex and trait vividness were entered, to control for these effects first. In the second step, flashforward emotionality and frequency (and also likelihood for depression factor only) were entered given the above significant correlations. We also explored moderation effects by groups in a third step, but these were not significant, so we omit them here for simplicity and due to their exploratory nature.
Table 4. Two-step linear regression analyses with anxiety or depression scores as dependent variables
n=82 (due to missing data on trait vividness; autism=3, maltreatment=1, typically developing=2); *p<.05, **p<.01, ***p<.001.
For the anxiety factor, the first model was significant, but no individual predictors were significant. The second model also including flashforward characteristics was also significant. Flashforward variables explained an additional 15% of variance. In this final adjusted model, flashforward emotionality was a significant positive predictor, meaning that the more emotional a self-relevant flashforward image was, the more anxiety symptoms were reported (Fig. 1). Trait imagery vividness was instead a significant negative predictor, meaning that better ability to generate vivid imagery of daily items was linked to less anxiety. Restricting the analyses to adolescents reporting positive images only (56.3% of the sample, n=49), the models were no longer significant (likely owing to loss in power).
Figure 1. Association between Flashforward Emotionality and Anxiety, Overall and by Groups.
Note. TD = typically developing; anxiety = factor score combining self- and caregiver RCADS scores; overall regression line includes 95% CI.
For the depression factor, the first model was statistically significant, with sex and age emerging as significant predictors. Female sex and older age predicted more depressive symptoms. The second model including flashforward characteristics was also significant. Flashforward variables explained an additional 7% of the variance. In this adjusted model, sex and age were no longer significant predictors. Instead, only trait imagery vividness was a significant negative predictor. That is, better ability to generate vivid imagery of daily items was linked to less depression. Restricting the analyses to adolescents reporting positive images only, both models remained significant but not any of the previous predictors (again likely owing to loss in power). However, flashforward likelihood emerged as a negative predictor, meaning that judging a positive flashforward as more likely to occur in real life was linked to less depression.
Exploring candidate mechanisms
Across groups, avoidant coping (measured with the CRIES; see Table 1) was significantly and positively correlated with flashforward emotionality, r(84)=0.30, p=.005, and frequency, r(84)=0.36, p<.001, but not clarity/likelihood, r(84)<0.13, p>.251. Contrary to the hypothesis, these correlations were not significant when restricting the analyses to the maltreatment group only, r(28)=0.05–0.31, p>.05.
Executive functioning (measured with the DEX-C; see Table 1) was not significantly correlated with any flashforward characteristics, r(84)<0.17, p>.332. However, when restricting analyses to the autism group only, and consistent with the hypothesis, worse executive functioning was correlated with increased flashforward frequency (controlling for sex and age), r(24)=–0.40, p=.045. This correlation in the autism group was significantly bigger than in the maltreatment group, r(22)=0.14, p=.506 [Z=1.78, p=.037], but not than in the typically developing group, r(25)=–0.40, p=.040 [Z=1.02, p=.155].
Discussion
We set out to explore involuntary flashforward imagery, and its relevance for mental health, in a sample of adolescents aged 10–16. Our preliminary work extends previous findings by considering a sample enriched with developmental differences with high psychopathology vulnerabilities, which partly reflected treatment-seeking young people within routine clinical services in the UK. Overall, we found links between flashforwards and anxiety which were applicable across the full sample, regardless of developmental backgrounds considered.
Unlike prior work looking at flashforwards in young people focusing on stressful events (e.g. Pile and Lau, Reference Pile and Lau2020) or worries (Shukla et al., Reference Shukla, Crew, Wu, Riddleston, Hutchinson, Kumari, Hughes and Lau2023), we did not restrict the valence of the mental image selected. Most adolescents selected a positive flashfowards image, consistent with emerging non-clinical studies (e.g. Barsics et al., Reference Barsics, Van der Linden and D’Argembeau2016), and which may function to maintain a positive self-image (Finnbogadóttir and Berntsen, Reference Finnbogadóttir and Berntsen2013). The most significant images were associated with career, education, and/or learning, which are key concerns of future orientation in adolescence (Nurmi, Reference Nurmi1991; Sawyer et al., Reference Sawyer, Afifi, Bearinger, Blakemore, Dick, Ezeh and Patton2012) and in line with imagery’s role in representing goals (Barsics et al., Reference Barsics, Van der Linden and D’Argembeau2016; Çili and Stopa, Reference Çili and Stopa2015). Those themes, albeit of mostly positive valence, are similar to the content of worries reported recently by young people during the COVID-19 pandemic (Shukla et al., Reference Shukla, Crew, Wu, Riddleston, Hutchinson, Kumari, Hughes and Lau2023). In the absence of direct instructions to focus on the negative, positive images appear to be more salient for young people when thinking about the future.
Contrary to the heightened flashforward hypothesis, there were no significant group differences on phenomenological characteristics of flashforwards between autistic/maltreatment-exposed and typically developing adolescents. This finding is in opposition to the presence of significant group differences on another form of future thinking assessed in the same sample. We previously found that the specificity of voluntarily generated future events is reduced in maltreatment-exposed compared with typically developing adolescents (Lau-Zhu et al., Reference Lau-Zhu, Chan, Gibson, Stark, Wang, Happé, Stacey and Cooper2024). Relatedly, we also did not find significant group differences in trait imagery vividness (of everyday scenes such as imagining certain sights and smells) in the current study (Andrade et al., Reference Andrade, May, Deeprose, Baugh and Ganis2013). Taken together, these add to the emerging picture that different aspects of emotional imagery dissociate, for instance between involuntary versus voluntary generation (Lau-Zhu et al., Reference Lau-Zhu, Henson and Holmes2019; Lau-Zhu et al., Reference Lau-Zhu, Henson and Holmes2021; Pearson and Westbrook, Reference Pearson and Westbrook2015) and subjective/emotional versus objective/cognitive aspects of imagery (Di Simplicio et al., Reference Di Simplicio, Renner, Blackwell, Mitchell, Stratford, Watson, Myers, Nobre, Lau-Zhu and Holmes2016; Di Simplicio et al., Reference Di Simplicio, Lau-Zhu, Meluken, Taylor, Kessing, Vinberg, Holmes and Miskowiak2019). It is still plausible that the heightened flashforward hypothesis is specific to negative rather than positive flashforwards, so measuring both separately would be an important next step. Future replications and extensions could also benefit from employing pre-registration, the lack of which is a limitation of our present study.
Flashforward emotionality emerged as the most consistent phenomenological characteristic predicting anxiety symptoms, after controlling for key covariates (including demographics and trait imagery vividness). Flashforward emotionality was not a significant predictor of depression, but this effect may become significant with a bigger sample. Nevertheless, the flashforward–anxiety association may still be bigger than a flashforward–depression one, which could be tested in future. With this caveat in mind, the current flashforward–anxiety association can be interpreted in several ways.
First, from a neuroscientific perspective, the evolved function of mental imagery is to predict threats (Schacter et al., Reference Schacter, Benoit and Szpunar2017), suggesting its central role in the over-estimation of threats in clinical conditions. Indeed, intrusive flashforwards are prominent in anxiety disorders and a focus of treatment innovation (Brewin et al., Reference Brewin, Gregory, Lipton and Burgess2010).
Second, considering that the emotionality rating captured both negative and positive images, it is plausible that imagery amplifies emotions of both valences, fuelling affect lability. For instance, in mania, positive images can become too intense and dysfunctional (Ivins et al., Reference Ivins, Di Simplicio, Close, Goodwin and Holmes2014). Affective lability, closely tied to anxiety (Di Simplicio et al., Reference Di Simplicio, Renner, Blackwell, Mitchell, Stratford, Watson, Myers, Nobre, Lau-Zhu and Holmes2016; Di Simplicio et al., Reference Di Simplicio, Lau-Zhu, Meluken, Taylor, Kessing, Vinberg, Holmes and Miskowiak2019; Lau-Zhu et al., Reference Lau-Zhu, Tuxen, Roerne and Di Simplicio2023), is a common feature of autism (Mazefsky et al., Reference Mazefsky, Herrington, Siegel, Scarpa, Maddox, Scahill and White2013) and maltreatment (Dvir et al., Reference Dvir, Ford, Hill and Frazier2014), both part of the developmental profiles of two-thirds of our sample (see Table 1). Thus, our flashforward measure may have tapped into a general propensity for experiencing both involuntary positive and negative imagery, as with the original conceptualisation in adults (Deeprose and Holmes, Reference Deeprose and Holmes2010).
Third, imagery’s appraisal may be important. A seemingly positive image (the majority in this sample) could have been appraised negatively. For example, an adolescent could hold a ‘positive’ image of becoming a scientist but then appraising it negatively as a difficult or uncertain goal. Imagery valence does not always match with appraisal valence (Ghita et al., Reference Ghita, Tooley and Lawrence2021). Negative appraisals may have been more common given the COVID-19 context.
Measuring affect lability and appraisals could clarify the different possibilities above in future studies. Development of a youth-specific flashforward measure with these considerations could also be fruitful. Longitudinal designs using cohort research or experience sampling (Thunnissen et al., Reference Thunnissen, aan het Rot, van den Hoofdakker and Nauta2022) could provide converging evidence to our preliminary cross-sectional findings.
The absence of moderation effects by group implies that the emotion amplification effects of flashforwards are applicable to adolescents across various developmental differences. Our data, and recent work (Ozsivadjian et al., Reference Ozsivadjian, Hollocks, Southcott, Absoud and Holmes2017), challenge long-held assumptions (often perpetuated in clinical settings) that all autistic individuals are poor imagers (Dance et al., Reference Dance, Jaquiery, Eagleman, Porteous, Zeman and Simner2021) or have limited imagination (Baron-Cohen et al., Reference Baron-Cohen, Wheelwright, Skinner, Martin and Clubley2001). The inherent heterogeneity within autism (Happé and Frith, Reference Happé and Frith2020) suggests that both weak and strong imagers can exist in the autism community. Individuals with mild-to-moderate intellectual disabilities (with or without autism) are also able to engage with mental imagery with appropriate support (Hewitt et al., Reference Hewitt, Steel, Hales, Hayden, Gundeslioglu, Tapp and Langdon2022; Hewitt et al., Reference Hewitt, Langdon, Hales and Larkin2023). Focusing on (flashforward) mental imagery, for whom it is relevant, could provide an important route towards therapeutic innovations to tackle the autism mental health challenge – a community priority (Weston et al., Reference Weston, Hodgekins and Langdon2016).
Mechanistically, flashforward frequency appeared to be related to key candidate processes. Higher frequency was associated with more avoidant coping, in line with cognitive theory highlighting the counterproductive effects of avoidance on memory intrusions (Williams et al., Reference Williams, Barnhofer, Crane, Hermans, Raes, Watkins and Dalgleish2007), and here extended to future-based intrusions. Higher flashforward frequency was also associated with less executive functioning, specifically in the autism group (O’Hearn et al., Reference O’Hearn, Asato, Ordaz and Luna2008), which can translate to detrimental effects over the control of flashforwards. Despite the absence of group differences in flashforward phenomenology (e.g. frequency or clarity), the underlying mechanisms may differ between groups. These analyses were exploratory but the obtained effect sizes pave the way for future mechanistic studies.
We also found that higher likelihood ratings on positive images were associated with fewer depressive symptoms, consistent with recent non-clinical studies (Beaty et al., Reference Beaty, Seli and Schacter2019; Ji et al., Reference Ji, Holmes, MacLeod and Murphy2019). An overly intense positive image (of a highly desired goal such as of future career) may lead to anxiety-linked dysregulation, but if it is underactive then it may confer risk for depression. It would be important to distinguish between the presence of unhelpful positive images from the absence of functional positive images, and to explore whether these factors distinctly influence mental health.
Clinically, enquiring about both negative and positive flashforwards more explicitly will be crucial in future, as these represent different intervention targets. Several imagery-based techniques have shown benefits in managing and/or reducing intrusive mental imagery in adults, such as imagery rescripting or metacognitive strategies aimed at changing the meaning or power of dysfunctional images, regardless of their valence (Hackmann et al., Reference Hackmann, Bennett-Levy and Holmes2011; Lau-Zhu et al., Reference Lau-Zhu, Tuxen, Roerne and Di Simplicio2023; Stopa, Reference Stopa2009). Recently, these techniques have been successfully applied in young people (Di Simplicio et al., Reference Di Simplicio, Appiah-Kusi, Wilkinson, Watson, Meiser-Stedman, Kavanagh and Holmes2020; Lau-Zhu et al., Reference Lau-Zhu, Farrington and Bissessar2022; Pile et al., Reference Pile, Williamson, Saunders, Holmes and Lau2021). A particularly promising approach leverages simple, imagery-interfering tasks, which rely relatively less on language-based skills (Lau-Zhu et al., Reference Lau-Zhu, Holmes, Butterfield and Holmes2017; Lau-Zhu et al., Reference Lau-Zhu, Henson and Holmes2019; Lau-Zhu et al., Reference Lau-Zhu, Henson and Holmes2021; Lau-Zhu et al., Reference Lau-Zhu, Farrington and Bissessar2022; Rackham and Lau-Zhu, Reference Rackham and Lau-Zhu2021). Given the potential benefits on depression, functional forms of spontaneous positive images could also be promoted (Blackwell et al., Reference Blackwell, Dooley, Würtz, Woud and Margraf2020).
Our analyses revealed the presence of suppressor effects (McCurdy et al., Reference McCurdy, Scozzafava, Bradley, Matlow, Weems and Carrion2023). Initially, trait imagery vividness (while controlling for age and sex) was not a significant predictor of anxiety/depression, but then became a significant negative predictor, when additional flashforward characteristics were controlled for (Table 4). This contrasts with a facet of flashforward imagery (i.e. its emotionality) as significant positive predictor of anxiety, demonstrating imagery dissociations. The trait imagery effect suggests the possibility that individuals who struggle with generating (and possibly flexibly manipulating) a variety of vivid mental scenes experience difficulties in downregulating affect lability, reflected in increased anxiety/depression symptoms. Overall, the relationship between different facets of imagery experience (across emotional and non-emotional domains) appears to be complex. Future investigations should consider these facets together rather than in isolation.
To conclude, the significance of future thinking in adolescent development is increasingly recognised, in particular its role in psychopathology. While there has been growing theoretical attention to involuntary mental imagery in emotional disorders, little exploration has been conducted in adolescents, especially among those with high psychopathology vulnerabilities due to developmental differences. Our preliminary work suggests that flashforward imagery holds relevance for understanding and improving mental health across adolescents with varying developmental profiles, including those related to autism and maltreatment. This approach is especially pertinent for anxiety problems, the incidence of which peaks in youth and where concerns about the future are central, often taking the form of imagery-rich cognitions. Further research is warranted to shed additional light on developmental, maintenance and intervention mechanisms.
Data availability statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.
Acknowledgements
We would like to thank the sponsoring and supporting organisations for facilitating this work including advice on design and recruitment strategy (with key individuals highlighted): Berkshire Child and Adolescent Mental Health Services (Dr Alice Farrington); Oxfordshire County Council ATTACH (Dr Claire Holdaway); Parents Talking Asperger’s (Karen Irvani); SAFE! Support for Young Victims (Chloe Purcell); St Frideswides Primary School in Oxford (Zoe Goodin); Autism Family Support Oxford; Autistica; and MQ Charity. We are particularly grateful for participating young people and families for their time and enthusiasm for completing this study through the challenging time of a global pandemic.
Author contributions
Alex Lau-Zhu: Conceptualization (lead), Data curation (lead), Formal analysis (lead), Funding acquisition (lead), Investigation (lead), Methodology (lead), Project administration (lead), Writing – original draft (lead), Writing – review & editing (lead); James Stacey: Conceptualization (supporting), Supervision (supporting), Writing – review & editing (supporting); Daisy Gibson: Formal analysis (supporting), Investigation (supporting), Writing – review & editing (supporting); Carmen Chan: Conceptualization (supporting), Supervision (supporting), Writing – review & editing (supporting); Myra Cooper: Conceptualization (supporting), Supervision (supporting), Writing – review & editing (supporting).
Financial support
This work was supported by funding for professional clinical psychology training from Oxford Health NHS Foundation Trust awarded to A.L.-Z. A.L.-Z. also reports support from a Medical Research Council Clinician Scientist Fellowship (MR/Y009460/1), John Fell Fund from the Oxford University Press (0014097), and National Institute of Health Research Imperial Biomedical Research Centre hosted at Imperial College Healthcare NHS Trust (PSP955).
Competing interests
The authors declare none.
Ethical standards
Ethical approval was obtained from the West Midlands–Solihull Research Ethics Committee (21/WM/0125). Informed consent to participate and for results to be published have been obtained. The Ethical Principles of Psychologists and Code of Conduct as set out by the British Association of Behavioural and Cognitive Psychotherapies and the British Psychological Society were abided by.
Open access
Comments
No Comments have been published for this article.