Neural mechanisms of time-based prospective memory: evidence for transient monitoring

Abstract

In daily life, we often need to remember to perform an action after, or at, a specific period of time (e.g., take pizza out of oven in 15 minutes). Surprisingly, little is known about the neural mechanisms that support this form of memory, termed time-based prospective memory (PM). Here we pioneer an fMRI paradigm that enables examination of both sustained and transient processes engaged during time-based PM. Participants were scanned while performing a demanding on-going task (n-back working memory), with and without an additional time-based PM demand. During the PM condition participants could access a hidden clock with a specific button-press response, while in the control condition, pseudo-clocks randomly appeared and were removed via the same response. Analyses tested for sustained activation associated with the PM condition, and also transient activation associated with clock-checks and the PM target response. Contrary to prior findings with event-based PM (i.e., remembering to perform a future action when a specific event occurs), no sustained PM-related activity was observed in anterior prefrontal cortex (aPFC) or elsewhere in the brain; instead, transient clock-related activity was observed in this region. Critically, the activation was anticipatory, increasing before clock-check responses. Anticipatory activity prior to the PM target response was weaker in aPFC, but strong in pre-Supplementary Motor Area (pre-SMA; relative to clock-check responses), suggesting a functional double dissociation related to volitional decision-making. Together, the results suggest that aPFC-activity dynamics during time-based PM reflect a distinct transient monitoring process, enabling integration of the PM intention with current temporal information to facilitate scheduling of upcoming PM-related actions.

Figure 1. Task Design.

Participants performed the 2-back condition of the n-back working memory task in either a control condition (Panel A) or time-based PM condition (Panel B), in counterbalanced order. The 2-back task presented a continuous sequence of words at a regular rate (1.5s) with target responses required when the current item matched that presented 2 trials back (arrow labeled “2-back”). In the control condition, at intermittent times, a non-informative clock randomly appeared (red font) with the instruction “press 3” (indicating the relevant button to press on the response box). Upon reacting to this prompt the clock would disappear. (arrow labeled “Control Prompt”). In the time-based PM condition, participants self-initiated a clock-check with the same button-press response (arrow labeled “participant clock-check”), leading to the appearance of an informative clock indicating the elapsed time since the start of the task block. They were asked to use this information to make a target response after 3 or 4 minutes. The 2-back task continued for an additional 20 seconds following the designated target time.

Figure 2. Clock-check responses across time

. The mean number of clock-checks in each minute during TBPM condition is displayed (referenced to the target time; e.g., Target-1 refers to last minute prior to target time). Clock-checks reliably increased in frequency with increasing proximity to the PM target time. Clock-checks in the final minute (Target-1) were significantly greater than all preceding minutes combined, consistent with prior findings, indicated by a * over the T-1 period (p<.05). ⁺ Target-4 only contains data from 4-minute trials. 3-minute trials were shifted accordingly to align the last minutes.

Figure 3. Block-related activation associated with on-going 2-back task.

Anatomical location of regions showing significant block-related activity in both Control and time-based PM conditions identified through a conjunction analysis. Color scale indicates minimum z-value across the two conditions. Prominent foci of activity are seen in lateral and medial prefrontal and parietal cortex, consistent with prior n-back findings.

Figure 4. Peri-event timecourses of clock-related activation.

Timecourses refer to 14-TR epochs (35 seconds) centered at the actual event onset time (TR-7, Blue Dashed Vertical Bar) A. Bilateral aPFC ROI (anatomical location shown in inset), exhibits a reliable increase in activation for time-based PM clock-checks (Purple) relative to Control clock-activation (Green). Two peaks of activation are present, one preceding clock onset (vertical Red dashed line at TR-9, adjusted to indicate hemodynamic response lag) and the other following it. The horizontal bar indicates that a significant PM > Control pattern was observed during the anticipatory, pre-clock period (p<.05). B. Control region in posterior parietal cortex (anatomical location shown in inset) that showed no difference in post-clock activation between time-based PM and Control conditions, as well as no pre-clock activation in either condition.

Figure 5. Peri-event timecourses of PM-response activation.

Timecourses refer to 14-TR epochs (35 seconds) centered at the time of PM-response/clock-onset (TR-7, blue dashed vertical bar) A. Right aPFC ROI (location shown in inset), exhibiting weaker activation for time-based PM responses (navy) relative to clock-checks (purple). Activity is reduced and delayed both prior to the PM response (TRs 7–9) and following it (TRs 10–14). The horizontal bar indicates the period in which a trend-level significant (p = .07) condition x time interaction was observed. B. Contrasting pattern observed in pre-SMA (location shown in inset), with greater activation for PM responses relative to clock-checks. The horizontal bar indicates the period over which a significant condition x time interaction was observed (p<.05).