Introduction

Mental time travel (MTT) refers to the ability to mentally project oneself to the past or to the future1,2. This capacity not only facilitates learning from past events but also fuels the anticipation of the future consequences of one’s choices, thereby influencing decision-making3 and psychological well-being4,5. Past and future MTT are associated with a shared network of brain regions, including the medial temporal lobes (MTLs), inferior frontal cortex, ventromedial prefrontal cortex (vmPFC), posterior cingulate, and lateral parietal regions6,7,8,9. Neuropsychological studies have particularly emphasized the role of the MTLs and the vmPFC in past and future MTT. Both patients with vmPFC and hippocampal damage exhibit impairments in constructing past and future events10,11,12,13. Notably, whereas in event construction task hippocampal patients primarily fail to provide spatial details and produce spatially fragmented scenes12,14, vmPFC patients additionally fail in providing multiple types of details, including the entities present and their sensory qualities15,16, suggesting that vmPFC plays a more basic, upstream role in event construction17,18. McCormick et al.14 proposed, therefore, that vmPFC may initiate the activation of higher-order memory structures in the neocortex (e.g., lifetime periods, self or event schemata) driving the collection of relevant details that the hippocampus then integrates into coherent scenes. It has been found, indeed, that the construction of both episodic memories and future thoughts is typically a generative process in which knowledge structures at different levels of specificity are progressively and transitorily accessed19,20. In particular, individuals first access high order, abstract memory structures such as general semantic memory (e.g., childhood) and personal semantic memory (e.g., my kindergarten) to then converge on lower-level memory structures such as specific events (e.g., that time I hid my teacher’s shoe)19,20. Activating high level semantic structures also helps determining the timing and relative order of past and future events during MTT21,22. Indeed, individuals often rely on reconstructive strategies to place events in time (e.g., this was before I got married)23. Even though the vmPFC has been extensively associated with schematic knowledge24, there have been few attempts to link this region to the activation of other types of semantic structures during MTT14,20,25, or to understand whether it plays a similar role during past and future MTT. This is in part due to the type of memory tasks with which MTT has been mostly assessed. MTT paradigms typically require participants to narrate specific personal past and future events, and therefore participants tend to underreport on more general memory structures, such as general and personal semantics. Indeed, semantic details are typically much fewer than episodic details when MTT reports are quantified with the Autobiographical Interview method26. Note also that the richness of constructed experience (e.g., number of details) is a sensitive but unspecific index of MTT performance. It is the output of a multiplicity of processes operating in concert to shape MTT. These include accessing general and personal semantic knowledge (e.g., lifetime periods), as well as computing temporal distances and orienting oneself in subjective time. These processes may be differentially engaged during past and future MTT and differentially affected by vmPFC damage.

Several authors have adopted a different approach to investigate MTT by using tasks that do not rely on linguistic fluency and tap self-projection and temporal orienting27,28,29,30,31. In an fMRI study, Arzy et al.9 asked participants to self-project either 8 years in the past or in the future (as opposed to stay in the present time) and presented famous faces manipulated to appear either younger or older or events. Participants had to judge whether each face or event laid in the past or future with respect to their currently adopted time location. Regardless of the stimulus being used, the results revealed that self-projection recruited a common brain network including the inferior frontal, anteromedial temporal, temporo-parietal, posterior parietal and insular cortices. Ciaramelli et al.32 recently used a similar approach to study the role of vmPFC in MTT. Participants are asked to project themselves either 10 years in the past or 10 years in the future and then they are presented with a series of events and asked to judge whether each event is past or future with respect to their currently adopted time location. The authors found that vmPFC patients, compared to healthy and brain damaged controls, have a deficit in self-projecting towards the future but not the past and in orienting themselves with respect to future but not past events. The asymmetry in the role of vmPFC in future vs. past MTT is compatible with the hypothesis that vmPFC mediates the activation of semantic structures during MTT14,20. Indeed, imagining the future relies more on schema-based knowledge than remembering the past, because we have no direct experience of future events22,33,34. However, the Ciaramelli et al.32’s study investigated whether vmPFC patients could place events correctly in the (relative) past and future, but not whether they would put the events in the correct chronological order within the past or the future. Moreover, that study involved specific personal and public past and future events rather than probing general memory structures, and therefore the findings may reflect, at least in part, a poor ability, on the vmPFC patients’ part, to conceive, or represent, some of the specific future events (e.g., first flying car; world peace).

Here, we tested whether vmPFC is necessary for MTT using a novel task that requires representing in time and orienting towards very common life events35. Participants were shown faces of individuals of various ages alongside short sentences about common life events (e.g., celebrating your grandchild’s birth). In the Past Projection condition, upon seeing a face and an event, participants had to say whether or not it was likely that the portrayed person had experienced the event 10 years earlier; in the Future Projection condition, they had to say whether or not it was likely that the portrayed person would experience the event in 10 years. This task involves mapping common events on a timeline of life, based on high order semantic knowledge about the temporality of life events, estimate the age of the portrayed person, orient in time 10 years in the past or in the future, and judge the likelihood of the events in the past or in the future.

To further study the ability of vmPFC patients to represent the temporality of life events, in a second study we explicitly required them to place a series of culturally-shared events (e.g., marriage, retirement) on a line demarcated according to age21,36. If vmPFC plays a crucial role in MTT, by activating general knowledge about life events and orienting in time (10 years in the past or in the future), then vmPFC patients’ probability judgments in the MTT task should not be modulated coherently as a function of the perceived age of the portrayed faces and the past vs. future projection condition. A deficit in activating general knowledge about common events should also result in an impairment in placing events appropriately on a timeline. Considering the prominent role played by the vmPFC in future-oriented cognition10,32,37,38,39, we expect that deficits in MTT would be more marked in the future compared to the past projection condition, and that deficits in the temporal ordering of life events would be more marked for relative future compared to relative past events.

Methods

Participants

Participants (N = 50) included 22 patients with focal brain lesions and 28 healthy controls. To establish the sample size, a power analysis was conducted using the PANGEA software (https://jakewestfall.shinyapps.io/pangea/). Considering the interaction of interest Age × Projection × Group, the power analysis suggested a total sample of 48 participants to detect a medium standardized effect size d ≥ 0.45, with power 1−β = 0.95. Patients were recruited at the Centre for Studies and Research in Cognitive Neuroscience (Cesena, Italy) and at the Istituti Clinici Scientifici Maugeri IRCCS (Castel Goffredo, Italy), based on their lesion site documented by magnetic resonance imaging (MRI) or computerized tomography (CT) scans (see Table 1 for patients’ demographic and clinical data). Inclusion criteria included an age comprised between 40 and 80, the presence of focal brain lesions, and passing the training phase of MTT task (≥ 75% correct answers; see below).

Table 1 Patients’ demographic and clinical data. TPL = Time post lesion (in years); aetiology = haemorrhagic (H), tumour (T), arteriovenous fistula (F), arteriovenous malformation (AVM), ischemic (I); ACE-III - Addenbrooke’s cognitive examination III40 (cut-off = 72); Raven’s progressive matrices, black/white version42 (cut-off ≥ 15); attentional matrices42 (cut-off ≥ 30); digit Span - Forward and backward43 (cut-off ≥ 3.75); corsi test - Forward and backward42 (cut-off ≥ 3.75); phonemic fluency (cut-off ≥ 17) and semantic fluency44 (cut-off ≥ 25); Stroop’s test45 - Errors (cut-off ≤ 4.24), interference (cut-off ≤ 36.92); TOL46 = Tower of London - Total moves (cut-off ≤ 70), rule violation (cut-off ≤ 70); WCST47 = Wisconsin card sorting test - Errors, perseverations (cut-off ≥ 10th percentile). An impaired performance is indicated by an *.
Full size table

Eleven patients had lesions centered on the vmPFC (vmPFC patients; 6 females; mean age = 59.45 years, s.d. = 9.62; mean education = 10.82 years, s.d. = 2.36), and eleven patients had lesions outside the vmPFC, forming the control patients group (6 females; mean age = 56.09 years, s.d. = 9.84; mean education = 13.27 years, s.d. = 4.34). The healthy control group included 28 participants (17 females; mean age = 60.39 years, s.d. = 6.07; mean education = 13.43 years, s.d. = 3.28) with no self-reported history of neurological or psychiatric disorders. The three groups were comparable in age [F2,47 = 1.194; p = 0.312; η2p = 0.048] and education [F2,47 = 2.492; p = 0.094; η2p = 0.096].

All participants provided written informed consent to take part in the study, which was approved by the local Ethics Committee (protocol n° 0023926; Department of Psychology, University of Bologna; protocol n° CE2717 Istituti Clinici Scientifici Maugeri) and conformed to the ethical standards of the Declaration of Helsinki (2013).

Neuropsychological assessment

General cognitive functioning, as assessed with the Addenbrooke’s Cognitive Examination-III (ACE-III)40,41, was preserved in the three groups (> 72/100 across groups). The patient groups also underwent an extended neuropsychological evaluation (see Table 1). Overall, attentional skills (Raven’s Standard Matrices and Attentional Matrices42), verbal and spatial short-term memory (Digit Span43, Corsi test42), and language and executive abilities (Phonemic and Semantic Fluency44, Stroop’s Test45, Tower of London46, Wisconsin Card Sorting Test47) were preserved in both groups. No significant differences were found between patient groups across tests (all ps ≥ 0.091).

Lesion mapping

Patients’ individual lesions, based on the most recent MRI or CT scans, were manually drawn by a trained neuroscientist directly onto each slice of the normalized T1-weighted template MRI scan from the Montreal Neurological Institute, using MRIcro48. Figure 1 shows the extent and overlap of brain lesions in vmPFC patients. Brodmann’s areas (BA) mainly affected were BA 10, BA 11, BA 25 and BA 32. The region of maximal lesion overlap occurred in BA 11 (mean = 10.69 cc, s.d. = 8.79), BA 10 (mean = 11.13 cc, s.d. = 10.94), BA 32 (mean = 5.83 cc, s.d. = 6.15), and BA 25 (mean = 1.62 cc, s,d, = 1.92). In control patients, the lesion involved a large region extending from the inferior frontal gyrus to the temporal and occipito-parietal cortex. Brodmann’s areas (BA) mainly affected were BA 38 (mean = 1.72 cc, s.d. = 1.97), BA 37 (mean = 4.47 cc, s.d. = 6.60), BA 48 (mean = 10.66 cc, s.d. = 14.63), BA 45 (mean = 5.09 cc, s.d. = 5.22). The two patient groups did not differ in lesion volume (t20 = − 0.146; p = .885).

Fig. 1
figure 1

Location and overlap of vmPFC patients’ brain lesions. The panel shows the lesions of the 11 patients with vmPFC damage projected on the same seven axial slices and on the sagittal view of the standard Montreal Neurological Institute brain. The level of the axial slices is indicated by white horizontal lines on the mesial view of the brain and by z-coordinates. The color bar indicates the number of overlapping lesions. Maximal overlap occurs in BAs 11, 10, 32, and 25. In axial slices, the left hemisphere is on the left side.

Full size image

Stimuli

We selected 5 male face identities (age range between 45 and 50 years) with neutral expressions from the Chicago Face Database49. Using FaceApp (https://faceapp.com/app), we digitally altered these faces to depict them as younger and older. The GNU Image Manipulation Program (GIMP; https://www.gimp.org/) was used to perform morphing between the rejuvenated or aged faces and their original counterparts. The percentage of morphing was modulated to obtain 5 desired age gradients: 40, 50, 60, 70, and 80 years (Fig. 2a). The perceived ages of the faces were confirmed through a pilot study. Four face identities were used for the main MTT task, and one for the instruction presentation and the training phase.

Experimental procedure

Participants sat in front of a PC screen (15.6″, resolution 1920 × 1080 pixels) at about 50 cm distance. The procedure included a task designed to investigate mental time travel (MTT task; Fig. 2), a control task to assess participants’ ability to estimate the age of the presented faces (Age Estimation control task), and a ‘Lifeline task’ requiring participants to place typical life events on a line representing the duration of one’s life (Fig. 3), which were performed in this order. The experimental session lasted approximately 90 min.

Fig. 2
figure 2

A schematic representation of the mental time travel task. In the upper panel (a), the figure shows the Past Projection condition (on the left) and the Future Projection condition (on the right). In the lower panel (b) the timeline of a single trial of the MTT task is depicted.

Full size image
Fig. 3
figure 3

Lifeline task. The figure represents the Lifeline task: participants were instructed to place each event beside the age at which they believe it typically occurred. Based on the participants’ choice, for each event a coloured bar appears on the horizontal line representing the lifespan (from 0 to 90 years).

Full size image

MTT task

Using OpenSesame50, participants were presented with a brief sentence describing one of six life events that typically occur around the age of 60 (selected in a pilot study; i.e., celebrating the 25th wedding anniversary​, paying off the mortgage​, celebrating the birth of the grandchild​, attending the child’s wedding​, finding a retirement home for the parent, retiring). After a 1000 ms interval, a face was displayed in the center of the screen for 2000 ms (Fig. 2b). The stimulus lasted up to 4 s or until the participant responded. In the Past Projection condition, participants were encouraged to project themselves back in time and required to indicate whether it was likely or unlikely that the person depicted had experienced the portrayed event 10 years ago (for example, “Is it likely or unlikely that he has retired 10 years ago?”). In the Future Projection condition, participants were encouraged to project themselves forward in time and required to indicate whether it was likely or unlikely that the person depicted would experience the portrayed event in 10 years (for example, “Is it likely or unlikely that he will retire in 10 years?”)9,35. Notably, we used different events occurring around the age of 60 as stimuli, and we used faces of different ages and identities rather than simply providing the age of a hypothetical person to discourage merely arithmetic strategies in solving the task and encourage self-projection 10 years back in the past or ahead in the future. Faces of different ages have indeed been widely used and proven effective in probing MTT9,35.

Participants were instructed to respond as quickly and accurately as possible by pressing either the ‘A’ or ‘L’ key (counterbalanced). In the Past Projection condition, an increase in the proportion of Likely responses (peak) was expected for faces around 70 years, whereas in the Future Projection condition, an increase in the proportion of Likely responses (peak) was expected for faces around 50 years. The inter trial interval was 750 ms.

The six events were presented in separate and randomized blocks. Each block consisted of two sub-blocks, one for the Past and one for the Future Projection condition, presented in a counterbalanced order (Fig. 2a). Within each block and experimental condition, the identities and ages of the faces were presented in random order. The total number of trials was 240 (4 facial identities × 5 age gradients × 6 events × 2 temporal projections).

Prior to the experimental task, a training session was conducted in which feedback on the accuracy of each response was provided. The training was repeated until participants reached 75% accuracy, and were given a maximum of 5 training sessions to reach the accuracy threshold (as in35). Faces and events adopted in the instruction presentation and in the training session were not used in the experimental task.

Age estimation control task

Using OpenSesame50, participants were presented with the same faces adopted in the MTT task, one at a time and in random order. Participants were instructed to report, for each face, the estimated age using the numeric keypad. Each face was displayed for 2000 ms, and the response time limit was set at 4000 ms. The interstimulus interval was 750 ms.

Lifeline task

Using PsychoPy51, a randomized list of 8 life events was presented on the computer screen along with a horizontal line representing the lifespan (from 0 to 90 years, with 5-year intervals; Fig. 3). The events were taken from the literature on cultural life scripts and covered the entire lifespan21,36. Participants had to place each event beside the age at which they believe it typically occurred in the current Italian culture.

Data analysis

MTT task

For each participant, we calculated the percentage of Likely responses depending on the age of the face and the temporal projection condition. To investigate potential differences among groups in the MTT task, a generalized linear model (GLM) was adopted using IBM SPSS Statistics (Version 28.0.1.1) with Projection (Past, Future) and Age (40, 50, 60, 70, 80) as within-subject factors, and Group (HC = healthy controls, CP = control patients, vmPFC = vmPFC patients) as between-subject factor.

Additionally, to determine the ages at which participants most likely associated the events, we calculated the peak by fitting a Gaussian curve to the percentages of Likely responses as a function of the age of the face (ranging from 40 to 80 years). Using OriginPro software (Version 2023), we extracted the mean of the distribution for each participant, corresponding to the peak of the Gaussian curve for each Projection condition (see35). The Gaussian curves demonstrated a good fit for the data across groups (HC: Past Projection, mean R2 = 0.988, range = 0.941 − 0.999; Future Projection, mean R2 = 0.980, range = 0.777–0.999; vmPFC: Past Projection, mean R2 = 0.956, range = 0.688–0.999; Future Projection, mean R2 = 0.972, range = 0.811−0.999; CP: Past Projection, mean R2 = 0.967, range = 0.885−0.999; Future Projection: mean R2 = 0.942, range = 0.789−0.999). In the Past Projection condition, curve fitting was unsuccessful in a healthy control individual. In the Future Projection condition, curve fitting was unsuccessful in a control patient. Therefore, the subsequent analysis was conducted excluding those values. A GLM was then conducted on the peaks of the Gaussian curves, with Projection (Past, Future) as within-subject factor and Group (HC, CP, vmPFC) as between-subject factor.

Age Estimation control task

To evaluate whether the ability to attribute ages to faces was comparable across groups, a repeated measures ANOVA was conducted on the age attributed to faces with Age (40, 50, 60, 70, 80) as within-subject factor and Group (HC, CP, vmPFC) as between-subject factor.

Lifeline task

To assess whether the ability to estimate the ages associated with typical life events was comparable among groups, a GLM on the age associated with each event was performed with Event (Primary school, Puberty, First Job, Marriage, Career Peak, Grandchild birth, Retirement, Death) as within-subject factor and Group (HC, CP, vmPFC) as between-subject factor. Due to an error in PsychoPy, the responses to the Primary school and Death item in a vmPFC patient and to the Grandchild birth item in a control patient were not recorded.

In all analyses, the Holm-Bonferroni correction was applied to account for potential significant differences and the 95% of confidence interval (CI) was reported.

Results

MTT task: percentage of likely responses

As expected, the GLM on the percentage of Likely responses showed a significant effect of the factor Age (Wald χ2(4) = 133.117; p < 0.001) and a significant interaction Age × Projection (Wald χ2(4) = 508.041; p < .001). In the Past Projection condition, the probability was greater for faces aged 70 (59.46%, CI = [55.48%, 63.45%]) than for faces aged 40 (15.64%, CI = [11.66%, 19.62%]), 50 (24.30%, CI = [20.32%, 28.28%]), 60 (45.13%, CI = [41.15%, 49.11%]) and 80 (44.07%, CI = [40.08%, 48.05%]; all ps < 0.001). In the Future Projection condition, the probability was greater for faces aged 50 (61.38%, CI = [57.40%, 65.36%]) than for faces aged 40 (44.21%, CI = [40.23%, 48.20%]), 60 (49.24%, CI = [45.26%, 53.22%]), 70 (26.28%, CI = [22.30%, 30.27%]) and 80 (14.19%, CI = [10.21%, 18.17%]; all ps < 0.001). Additionally, the factor Group (Wald χ2(2) = 51.358; p < 0.001), the interactions Age × Group (Wald χ2(8) = 79.934; p < .001) and Projection × Group (Wald χ2(2) = 7.625; p = .022) were significant, and were qualified by a significant Age × Projection × Group interaction (Wald χ2(8) = 39.147; p < .001; Fig. 4). Post hoc comparisons showed that in the Past Projection condition vmPFC patients gave a higher percentage of Likely responses to faces aged 40 years (HC, p < .0.001; CP, p = .0.002) and a higher percentage of Likely responses to faces aged 50 compared to the two control groups (HC, p < 0.001; CP, p = 0.018), suggesting that they overestimated the probability that an event had already occurred at a relatively younger age. No differences were found between the two control groups (all ps > 0.365) or for other age gradients among vmPFC and the control groups (all ps > 0.138). In the Future Projection condition, vmPFC patients gave a higher percentage of Likely responses to faces aged 40 years compared to the two control groups (all ps < 0.001), which did not differ from each other (p > 0.999), suggesting that they overestimated the probability that an event will be already occurred at a relatively young age. Finally, the CP group reported a greater percentage of Likely responses to faces aged 80 years and 70 years compared to the HC group (80 years, p < 0.001; 70 years, p = 0.004) but not compared to the vmPFC group (80 years, p = 0.602; 70 years, p > 0.999). No differences were found for other age gradients among vmPFC and the control groups (all ps > 0.491) The factor Projection (p = 0.297) was not significant.

Fig. 4
figure 4

Mental time travel task results. The graph depicts the percentages of Likely responses observed in the three groups (HC = healthy control; CP = control patients; vmPFC = vmPFC patients) as a function of Age in the Past Projection (left) and Future Projection condition (right). Error bars represent 95% CI.

Full size image

MTT task: mean peak

The GLM on the peak values of the of the Gaussian curves showed a significant effect of Projection (Wald χ2(1) = 68.055; p < 0.001): across groups, the mean peak for the Past Projection (mean = 67 years, CI = [64, 69]) condition was significantly higher than that for the Future Projection condition (mean = 51 years, CI = [48, 54]), suggesting that all groups understood the task and projected themselves to the past or the future depending on the experimental condition. Crucially, the factor Group was significant (Wald χ2(2) = 11.207; p = 0.004): the vmPFC group exhibited a shift of the peaks of the psychometric curves toward a younger age (mean = 54 years, CI = [50, 58]) compared to both control groups (HC: mean = 60 years, CI = [58, 63], p  = 0.007; CP: mean = 62 years, CI = [58, 66], p = 0.007), which did not differ significantly from each other (p = 0.460). These findings indicate that vmPFC patients anticipated the expected age at which the portrayed individuals experienced the events, during both past and future MTT. The interaction Projection × Group was not significant (p = 0.502).

Age estimation task

The repeated measures ANOVA conducted on the responses from the Age Estimation task revealed a significant main effect of Age (F(2.530, 118.899) = 375.477, p < 0.001, η2p = 0.889; Fig. 5), indicating that estimated ages increased with the age of the presented faces, with ages of 40 years attaining lower estimates compared with faces of 50 years that in turn obtained lower estimates than faces of 60, 70, and 80 years (all ps < 0.001). The main effect of Group (p = .463) and the Group × Age interaction (p = 0.114) were not significant, suggesting that all participants were able to estimate the correct age of the portrayed faces.

Fig. 5
figure 5

Age Estimation task results. The graph shows the mean estimated ages provided by the three groups (HC = healthy control; CP = control patients; vmPFC = vmPFC patients) as a function of the veridical ages of the portrayed faces. Error bars represent 95% CI.

Full size image

To further examine group differences in age estimation accuracy, error scores were calculated by subtracting the veridical ages of the portrayed faces from the ages reported by the participants. A repeated-measures ANOVA was conducted on the error scores, with Age (40, 50, 60, 70, 80) as a within-subject factor and Group (HC, CP, vmPFC) as a between-subject factor. A significant main effect of Age was found (F(2.595, 121.987) = 26.742, p < 0.001): error scores for faces aged 40 (mean: +4 years) were significantly different from those for faces aged 50 (mean: +2 years), 60 (+ 2 years), 70 (− 1 year), and 80 (− 3 years; all ps < 0.014). Similarly, significant differences were observed between faces aged 50 and both 70 and 80 (all ps < 0.001), as well as between faces aged 60 and both 70 and 80 (all ps < 0.003). The Group factor (F(2, 47) = 0.720, p = 0.492) and the Age × Group interaction (F(5.191, 121.997) = 1.482, p = 0.198) were, however, not significant, meaning that participant groups were equally able to estimate age.

Lifeline task

The GLM conducted on judgments provided by participants at the Lifeline task showed a significant effect of the factor Event (Wald χ2(7) = 11138.913; p < 0.001), indicating that participants’ responses were modulated by the different events, with the event Primary school (mean = 6 years, CI = [5, 7]) attaining lower estimated ages than the event Puberty (mean = 13 years, CI = [12, 14]), in turn attaining lower estimated ages than the event First Job (mean = 24 years, CI = [22, 25]), Marriage (mean = 29 years, CI = [27, 30]), Career Peak (mean = 43 years, CI = [41, 44]), Grandchild birth (mean = 59 years, CI = [58, 60]), Retirement (mean = 65 years, CI = [63, 66]), and Death (mean = 83 years, CI = [82, 84]) (all ps < 0.001). The factor Group was significant (Wald χ2(2) = 22.768; p < 0.001): post-hoc comparisons indicated that vmPFC patients were likely to associate all events with a significantly younger age compared to both control groups (HC, p < 0.001; CP, p = 0.003; Fig. 6). The Group × Event interaction was not significant (p = 0.140).

Fig. 6
figure 6

Lifeline task results. The figure shows the mean judgment provided by each group (HC = healthy control; CP = control patients; vmPFC = vmPFC patients). Violin plots were adopted to show data distribution and the jittered raw data. The central line in each box display the median, the edges of the box indicate the interquartile range (IQR) and the whiskers extend to 1.5 times the IQR.

Full size image

Exploratory correlations

To test whether the MTT task and the Lifeline task rely on overlapping processes for estimating the timing of common events, we conducted Pearson’s correlations between the mean age associated with events in the Lifeline task and the peak values relative to the MTT task in the Past and Future Projection condition, which were significantly correlated between each other (r(48) = 0.445, p = 0.002). A significant positive correlation was found between the peak values for past MTT and the mean age of events in the Lifeline task (r(49) = 0.418, p = 0.003), such that participants who underestimated the age at which events occur in the Lifeline task were also those that did so during past MTT. By contrast, no correlation was observed between the peak values for future MTT and the mean age of events in the Lifeline task (r(49) = 0.044, p = 0.763), meaning that future MTT recruited (partially) different processes.

Discussion

The present study examined the role of vmPFC in the ability to activate higher-order (semantic) memories about life events during MTT. vmPFC patients, patients with lesions outside the vmPFC, and healthy individuals projected themselves into past and future moments to judge the likelihood with which individuals of different ages had experienced or were to experience typical life events (MTT task) and placed typical life events on a timeline representing the duration of life (Lifeline task).

The results in the MTT task showed that vmPFC patients, as well as the other groups, associated higher age peaks with activities in the Past compared to the Future Projection condition, suggesting they understood the task requirements and modulated their responses depending on the age of the portrayed individual and the temporal projection condition. However, vmPFC patients showed specific impairments in both the Past and Future Projection conditions. In the Past Projection condition, vmPFC patients gave more Likely responses in association with younger faces (40 and 50 years) compared to the control groups. This suggests that they anticipated the age at which events typically occur in the past. Analogously, in the Future Projection condition, vmPFC patients gave more Likely responses in association with younger faces (40 years) compared to the control groups, again anticipating the age at which events are expected to occur. Indeed, vmPFC patients systematically associated the peak probability of events with younger ages compared to the control groups in both temporal conditions, despite able to correctly estimate the age of the faces, suggesting they consistently anticipated the occurrence of events during MTT.

This interpretation is confirmed by an additional task where participants were asked to arrange typical lifetime events on a line representing the duration of life21. We found, again, that vmPFC patients systematically anticipated the age associated with typical life events compared to the control groups. The observed tendency to anticipate events in vmPFC patients parallels the findings from Rasmussen and Berntsen52 in Alzheimer’s disease (AD) patients, who were instructed to report on the most important events in a prototypical life and estimate the age at which each event would occur. AD patients underestimated the ages associated with events compared to healthy controls. The authors attributed the deficit to AD patients’ prefrontal dysfunction, which would impair processing of temporal schematic knowledge53.

Why would vmPFC patients systematically anticipate life events in time during MTT? One possibility is that vmPFC patients have a shortened time perspective, with a reduced representation of the future, and therefore tend to systematically “translate” events toward the past. Fellows and Farah54 asked vmPFC patients, patients with dorsolateral prefrontal cortex lesions and healthy controls to imagine events they could potentially experience in the future and to date them. Despite conceiving events similar in content to the other groups’, vmPFC patients imagined the events to occur sooner. If we conceptualize events along a mental timeline, with the past represented on the left and the future on the right27,55,56, a compressed right (future) side would push all events toward the left, leading to an anticipation of events. This interpretation also aligns with the study of Ciaramelli et al.32, in which vmPFC patients systematically misjudged events as past (as opposed to future), which again points to a relative anticipation of events in time. An alternative explanation stems from studies highlighting the role of vmPFC in encoding the value of rewards and the effort associated with actions57,58,59,60. For example, vmPFC activity increases as the difficulty of an action decreases58,60 and lesions to the vmPFC reduce the subjective value of future rewards37,39. vmPFC damage might lead to a devaluation of events, or a diminished perception of the effort required to achieve them. Considering that dating events rely on reconstructive processes23, devaluing events can lead to perceiving them as accomplishable in less time, underestimating the time associated to each event.

Some limitations of the present study should be noted. First of all, the strategy with which individual participants accomplished the MTT task is not clear. Although we took precautions that encouraged mental time travel based on an appraisal of the age of individuals’ faces9,35, it cannot be excluded that some participants resorted to the “arithmetic” strategy of estimating the age of occurrence of an event, the age of the portrayed individual now and 10 year in the past or the future, and then make probability judgments. This possibility would introduce heterogeneity in our results, but would not undermine our central finding that vmPFC patients have an altered (anticipated) representation of common events in time. We, however, doubt that the MTT task was accomplished exclusively by timing relevant events, because in this case we should have found a correlation between the results (impairments) in the MTT task and in the Lifeline task, which requires placing events in time. This correlation was present with past MTT, but absent with future MTT, supporting the hypothesis that the MTT task likely required different (self-projection) processes.

An additional limitation of the study is that vmPFC is a large brain region but our sample size did not allow subregional analysis to investigate which sector of vmPFC was crucially associated with MTT. Additionally, it is possible that regions outside vmPFC also played some role in MTT. For instance, temporal and lateral frontal lesions in control patients could explain their difficulties in the Future Projection condition; we do not interpret them further as we did not have a priori hypotheses. Future studies could investigate whether vmPFC patients’ temporal underestimation deficit would extend to time perception tasks involving shorter (seconds) durations61,62.

For the time being, we have shown that vmPFC integrity is crucial for representing the temporality of past and future life events during MTT, and that vmPFC damage leads to a systematic anticipation of events in time. vmPFC may be necessary to mediate the activation of higher order memory structures, such as (future-oriented) temporal schemata, supporting the coherent orientation in subjective time during MTT.