The hippocampus dissociates present from past and future goals

Tracking personal goals requires monitoring what has been accomplished, what is currently prioritized, and what remains to be done, all varying by temporal distance from the present. Goals dynamically shift from future to present to past as time progresses, necessitating continual updating of their relevance. The hippocampus, central to episodic memory and mental time travel, exhibits functional specialization along its longitudinal axis, with posterior regions supporting more detailed representations and anterior regions more abstract, generalized ones. Prior observations in animals suggest hippocampal coding of time (e.g., time cells), raising the question of whether human hippocampal long-axis organization extends to temporal distance for goals. The study tests whether goals are mapped along the hippocampal long axis based on temporal proximity, with present goals engaging posterior hippocampus and temporally removed (past and future) goals engaging anterior hippocampus, while holding goal content constant.

Prior work identifies anatomical and functional specialization along the hippocampal long axis, with increasing representational granularity from anterior to posterior regions (e.g., Poppenk et al., Strange et al.). Rodent studies have revealed place and time cells, indicating hippocampal involvement in representing moments and sequences over time (e.g., MacDonald et al., Pastalkova et al.). Human neuroimaging and cognitive theories link the hippocampus to episodic memory, mental time travel, and a gradient of detail along the long axis. These literatures motivate examining whether the long-axis organization supports temporal mapping of goal relevance, distinguishing present from temporally remote (past/future) goals.

Design: Two-day experiment with training (Day 1) and ultra-high-field 7T fMRI scanning (Day 2). Participants completed a space mission-themed task spanning four ‘Mars years,’ repeatedly judging when specific goals needed to be accomplished relative to their current year. Identical goal stimuli recurred each year, but their temporal relevance shifted (distant future, near future, current, near past, distant past). Additional categories included always-needed and never-needed goals. Participants: 34 recruited; 31 included after exclusions (age M = 27.0, SD ≈ 4.3–4.5, range 19–36; n = 15 females). All provided informed consent; IRB-approved. Stimuli and training: 30 goals across six timeframes (year 1, year 2, year 3, year 4, always, never), five per timeframe. Categories included space suit maintenance, spacecraft maintenance, diet, exercise, and recreation. Participants learned goal–timeframe associations via view-and-quiz training. Stimuli valence, arousal, and familiarity were pre-rated online and balanced across sets. Task procedure (scanner): Across the four mission years, participants indicated for each goal whether it was applicable in the current year (CY), distant future (DF), near future (NF), near past (NP), distant past (DP), always (AN), or never (NN). Button options changed as years advanced (e.g., future options removed in the last year). Buttons rotated positions each trial to avoid motor confounds. The task was self-paced with intermittent fixation. Total trials: 120 (DF 15, NF 15, NP 15, DP 15, CY 20, AN 20, NN 20), distributed across years. Behavioral analyses: Reaction times (RTs) were log-transformed. Linear mixed-effects models (lme4/lmerTest in R 3.6.0) assessed effects of temporal condition (DF, NF, CY, NP, DP) with participant ID as a random effect and game year as fixed and random effects; outliers and incorrect trials removed (94.5% retained). A second model contrasted trials with versus without temporal computation (DF, NF, NP, DP vs CY, AN, NN). Post hoc comparisons used emmeans with Tukey adjustment; significance evaluated via Type II/III ANOVA with Satterthwaite df. MRI acquisition: 7T Siemens Magnetom, 32-channel head coil. T1-weighted MPRAGE (0.7 mm iso; TR 6000 ms; TE 5.14 ms; FoV 320×320). Functional multi-echo multi-band EPI (2.5 mm iso; TR 1850 ms; TE = [35–43] ms; MB=2; in-plane acceleration; flip 7°; FoV 640×640). Preprocessing: ME-ICA (multi-echo ICA) denoising to separate BOLD from non-BOLD components; subsequent FSL processing: skull stripping, boundary-based registration to T1 and standard space, trimming to task completion, 5 mm FWHM spatial smoothing to satisfy GRF assumptions. Neuroimaging analyses: ROI and whole-brain GLMs. Hippocampal ROIs derived via FreeSurfer subcortical segmentation; long-axis analyses contrasted temporally removed (remote: DF+NF+NP+DP) versus current (CY) conditions. Second-level mixed effects (FLAME). Multiple comparisons controlled using Gaussian random field theory with FWE voxel-wise correction at two-tailed p = 0.025. Exploratory whole-brain GLMs assessed broader patterns. Connectivity: generalized psychophysiological interaction (gPPI) using anterior/posterior hippocampal seeds to examine condition-dependent coupling.

Behavior: Participants’ log RTs differed by temporal condition (Type II ANOVA: F(4, ≈2236.3) = 64.83, p < 0.001). Post hoc comparisons showed current (CY) goals were processed faster than all temporally removed conditions (distant/near future, near/distant past). In a model including AN and NN, trials without temporal computation (CY, AN, NN) were faster than those with temporal computation (DF, NF, NP, DP); CY and AN did not differ significantly. Neuroimaging (hippocampus): Temporally removed (remote) goals elicited stronger activation than current goals in the left anterior hippocampus; current goals elicited stronger activation in the left posterior hippocampus. Reported foci (MNI): Current > Remote, left posterior hippocampus, Z = 4.28 at (-30, -31, -14.5) and Z = 3.75 at (-30, -36, -12). Remote > Current, left anterior hippocampus, Z = 4.48 at (-15, -8.5, -14.5); right anterior hippocampus also showed an effect, Z = 3.79 at (17.5, -21, -14.5). Effects were corrected for multiple comparisons (FWE voxel-wise, two-tailed p = 0.025). The contrast localization along the y-axis aligned with anterior (around y = -21) and posterior (y = -31 to -36) hippocampal boundaries. Whole brain: Temporally removed goals predominantly activated anterior brain regions, whereas current goals more strongly activated posterior regions, consistent with frontal involvement in planning/future retrieval. Connectivity: gPPI analyses (anterior/posterior hippocampal seeds) examined condition-dependent coupling patterns consistent with differential engagement for temporally removed versus current goals (details in Supplementary Tables).

The findings support the hypothesis that the hippocampal long axis differentiates goals by temporal distance from the present. Behaviorally, present-relevant goals are processed more quickly than temporally removed (past/future) goals, indicating distinct cognitive demands. Neurally, identical goals are dissociated within the hippocampus based on temporal relevance: anterior hippocampus preferentially represents temporally removed goals, while posterior hippocampus represents current goals. This extends long-axis specialization beyond spatial and episodic representational granularity to include temporal mapping of goal relevance. Whole-brain patterns further align temporally removed goals with anterior/frontal systems implicated in future-oriented cognition and planning, and current goals with more posterior regions. Together, results advance understanding of how the hippocampus anchors a temporal ‘timestamp’ to goals, contributing to models of time in memory and decision-making.

This study demonstrates that the human hippocampus organizes goals along its anterior–posterior axis according to temporal relevance: anterior hippocampus for temporally removed (past/future) goals and posterior hippocampus for current goals. By holding goal content constant and manipulating only temporal distance, the work provides evidence that the hippocampus maps goal timestamps. Future research should examine how multiple temporal representations are integrated within the hippocampus during memory and decision making, and how hippocampal–cortical interactions support dynamic updating of goal relevance over time.