Post-study caffeine administration enhances memory consolidation in humans

Many studies have documented the effects of caffeine as a cognitive enhancer¹. However, its effects on long-term memory have not been investigated in detail. The general consensus among past studies is that caffeine has little or no effect on long-term retention¹. However, caffeine has been always administered before learning; thus, effects on memory are impossible to dissociate from other effects of caffeine such as increased arousal, vigilance, attention and processing speed. We used a post-study design (drug administered after subjects have had an opportunity to study the material) based on animal studies², where effects of certain agents on memory consolidation are optimally detected after the learning experience.

We conducted a randomized, double-blind, placebo-controlled trial in caffeine-naive participants. On day 1, participants incidentally studied images of objects, then received either 200 mg of caffeine or placebo. We collected salivary samples at baseline and 1 h, 3 h and 24 h after administration of caffeine or placebo to quantify caffeine metabolites (Fig. 1a and Supplementary Fig. 1). Twenty-four hours after the study session, we evaluated participants’ recognition performance using some items they saw the previous day (targets), some new items (foils) and some items that were similar but not identical to ones they saw before (lures; Fig. 1a). Correctly identifying these lures as ‘similar’ has been previously shown to be associated with hippocampal activity³.

We found that participants in the group that received caffeine had a significant increase in caffeine metabolites (Supplementary Fig. 2) at the 1 h and 3 h time points, which then returned to baseline amounts over a 24-h washout period. Participants who received caffeine were more likely to call lure items ‘similar’ rather than ‘old’ compared to participants who received the placebo (Fig. 1b), whereas we found no group differences in rates of target hits (Fig. 1c) or foil rejection (Fig. 1d). D′ (z(target hits) – z(false alarms)) was not significantly different among groups (t₄₂ = 0.60, two-tailed P = 0.55), hence basic recognition memory was unaltered. We calculated a lure discrimination index (LDI) as P(‘similar’|lure) minus P(‘similar’|foil) to correct for response bias and found a significant difference between groups (t₄₂ = 2.0, two-tailed P = 0.05; Fig. 2a). This suggests that caffeine enhanced consolidation of the initial study session such that discrimination during retrieval was improved. Both groups had similar variance as assessed by Levene’s test for (D′) (P = 0.44) and LDI (P = 0.96).

To rule out any effects of caffeine on retrieval, we conducted a delayed manipulation, in which we administered caffeine 1 h before a test (24 h after the initial study session). We observed no significant enhancement compared to placebo, suggesting that caffeine does not affect any other aspect of retention performance (t₅₅ = 0.63, P = 0.53; Fig. 2a).

Next, we determined whether the effects of caffeine were consistent across lure similarity. In prior work, we had quantified the similarity of the stimuli based on the tendency of participants to erroneously produce false alarms⁴ and used that metric here to classify stimuli according to similarity. A two-way analysis of variance (ANOVA) revealed a significant main effect of similarity (F_2,88 = 12.87, P = 0.05) and a significant main effect of caffeine (F_1,42 = 4.07, P = 0.001) with a nonsignificant interaction (F_2,84 = 1.0, P = 0.37; Fig. 2b).

To determine whether there is an optimal dose range, we repeated the experiment with a placebo and different doses of caffeine (100 mg, 200 mg and 300 mg). We combined data across the two experiments for the placebo and 200 mg caffeine conditions to increase power. We found that performance for the 200 mg caffeine condition was higher than that for placebo (t₇₁ = 2.0, two-tailed P = 0.049; n = 35 subjects for 200 mg caffeine condition, n = 38 subjects for placebo condition; Fig. 2c). Also, performance for the 200 mg caffeine condition was higher than that for 100 mg (t₄₈ = 2.19, two-tailed P = 0.033), which was not significantly different from that for placebo (t₅₁ = 1.0, two-tailed P = 0.32). Finally, performance for the 300 mg caffeine condition was marginally higher than that for placebo (t₄₆ = 1.81, P = 0.07) but not different from the 200 mg caffeine condition (t₄₃ = 0.32, P = 0.75). We combined the higher doses together and compared them against the combination of placebo and 100 mg dose. The comparison was significant (t₉₆ = 2.77, P = 0.007). Thus, we conclude that a dose of at least 200 mg is required to observe the enhancing effect of caffeine on consolidation of memory.

Finally, we examined the relationship between the enhancement in performance and change in amounts of caffeine metabolites found in saliva to account for individual differences in metabolic function. We found a pattern similar to the dose-response results reported above (Fig. 2d). There was additional evidence of an inverted U-shaped dose-response curve. We compared the linear fit versus a second-degree polynomial (quadratic). The quadratic-curve fit R² was 0.81, whereas the linear fit R² was 0.45. A change F test demonstrated that the quadratic fit was significantly better than the linear fit (F = 29.5, P = 0.001).

Numerous studies in animals have shown that caffeine has neuroprotective effects^5–8. Prior work also found a positive effect of post-training administration of caffeine on consolidation of memory⁹. Notably, a recent study suggests that caffeine in floral nectar may boost memory for reward in honeybees¹⁰, suggesting that the mnemonic effects of caffeine may not be limited to mammals. No study to our knowledge has demonstrated a positive effect of caffeine on human long-term memory while excluding nonmnemonic effects. Our results demonstrate that caffeine enhanced consolidation once these effects are appropriately controlled.

Limitations of this study include subjects’ awareness of being involved in a caffeine study. To address this, we asked subjects whether they thought they were administered caffeine or placebo, and responses were distributed equally between perceived placebo and caffeine regardless of condition. Our final sample size for the 300 mg dose was small (n = 10 subjects) after eliminating subjects who did not conform to the protocol, thus the enhanced effect at the 300 mg dose could have been weakened by the smaller sample.

Although the mechanisms by which caffeine enhances memory consolidation remain largely unclear, there are several possibilities. First, by blocking adenosine, caffeine can prevent it from inhibiting norepinephrine, which has positive effects on consolidation of memory². We have previously demonstrated a relationship between norepinephrine and pattern separation¹¹. Another possibility is that caffeine’s action in the CA2 region of the hippocampus, which is highly enriched in adenosine A1 receptors¹², enhances long-term potentiation in this subfield¹³, which may have a role in certain types of memories¹⁴. In addition to the hippocampus, several other regions may have a role in consolidating memories, including the anterior cingulate cortex¹⁵ and the medial prefrontal cortex¹⁶. Other brain regions may have a role in modulating consolidation such as the basolateral amygdala^17,18 and mesolimbic dopaminergic areas¹⁹. Examining the contributions of these regions and the effects of caffeine on their processing is further crucial to understanding of psychostimulant-induced memory potentiation.

Future experiments should be conducted to understand the mechanisms by which caffeine can potentiate memory. Given the widespread use of caffeine and the growing interest in its effects both as a cognitive enhancer and as a neuroprotectant, these questions are of critical importance.