Caffeine enhances discrimination performance 24 h after study. (a) Outline of study design. After arrival of screened subjects, a baseline salivary sample was collected. Then the encoding task was administered. This was an incidental indoor-outdoor judgment task (stimuli every 2,500 ms, with an interstimulus interval (ISI) of 500 ms). After encoding, subjects were administered either 200 mg caffeine or placebo pills. After 1 h and 3 h, additional saliva samples were collected. Subjects returned 24 h later for testing. Before a recognition test, a final saliva sample was collected. Recognition was tested using an old-similar-new judgment task (stimuli every 2,500 ms with a 500-ms ISI) using targets, foils and similar lures that are particularly sensitive to hippocampal pattern separation. (b) Lure discrimination by subjects (i.e., whether subjects had a higher propensity to call lure items ‘similar’ rather than ‘old’) (t₄₂ = 1.79, one-tailed P = 0.04). *P < 0.05, one-tailed. (c,d) Target hit rates (c) and foil rejection rates (d) (t₄₂ = 0.59, one-tailed P = 0.27 and t₄₂ = 0.15, one-tailed P = 0.44 between groups that received caffeine and placebo, for data in c and d, respectively). Error bars, ±s.e.m.; n = 20 subjects (caffeine) and n = 24 subjects (placebo).

Impact of caffeine on consolidation and variable dose effects. (a) LDI in subjects administered placebo or caffeine immediately after the study session (immediate), or caffeine 24 h after the study session (delayed). LDI in immediate caffeine group was enhanced compared to placebo (t₇₁ = 2.0, two-tailed P = 0.049). LDI in the delayed group was no different from placebo (t₅₅ = 0.63, P = 0.53). (b) Analysis by item similarity showed a significant main effect of similarity (F_2,88 = 12.87, P = 0.001) as well as a main effect of caffeine (F_1,42 = 4.07, P = 0.05). (c) Discrimination as a function of indicated caffeine dose (200 mg caffeine compared to placebo, t₇₁ = 2.0, two-tailed P = 0.049 and compared to 100 mg caffeine, t₄₈ = 2.19, two-tailed P = 0.033). Discrimination at 200 mg and 300 mg caffeine doses compared to placebo and 100 mg caffeine combined, t₉₆ = 2.77, two-tailed P = 0.007. (d) Rebinned subject performance based on change in the amount of caffeine from baseline divided into quartiles (no change, low change, medium change and high change) showing evidence for an inverted U-shaped dose-response curve. The quadratic curve fit (R² = 0.81) was significantly better than the linear fit (R² = 0.45) for discrimination as a function of caffeine change (F = 29.5, P = 0.001). Error bars, ±s.e.m.*P < 0.05.