This research study aims to determine whether and how caffeine intake affects learning process through reward feedback compared to placebo. The data acquired from this study would improve our understanding on the consequence and mechanism of caffeine intake in the aspect of learning process.

Participants will perform a reinforcement learning task (i.e. Probabilistic Selection Task) and a motor inhibition task (i.e. Go/NoGo task) in a brain scan. The scan will be done with the Siemens Biograph mMR positron emission tomography (PET)/ magnetic resonance imaging (MRI) 3 Tesla scanner. The PET allows us to see the changes in the "reward signals" - dopamine - in the brain using a radioactive dye called [11C]Raclopride. The MRI, on the other hand, enables us to take detailed pictures of the brain activities during cognitive tasks using a high-powered magnet. Reviewing these pictures will help us understand the influence of caffeine on reward signals and brain activities during the learning process.

Adenosine 2A receptors (A2AR) colocalize with and exerts allosteric antagonism to dopamine D2 receptors (D2R) by co-forming functional heterodimers in the striatum. Preliminary studies using positron emission tomography (PET) with [11C]Raclopride have shown increased D2/D3R availability by A2AR antagonism with caffeine and decreased D2/D3R availability by enhanced adenosine signaling during sleep deprivation, supporting the notion of A2AR-D2R interactions in vivo. However, how A2AR-D2R interactions contribute to D2R-mediated neurocognitive functions is scarcely investigated.

Reinforcement learning, a dopamine-mediated cognitive process crucially involved in various human behaviors including habit, preference, belief, and resistance to change, is often found altered in dopamine-associated disorders. For instance, hyper-dopaminergic function in the striatum, as observed in psychosis, leads to a reduction in reward learning and a blunted task-related neural activity. Through the antagonistic effect of A2AR on D2R signaling, blocking A2AR can potentially enhance D2R-mediated negative reinforcement, a.k.a. a "no-go" response. In rodents, A2AR agonists diminish reinforcement of psychostimulants, while an A2AR antagonist can facilitate reward-seeking effects of reinforcers. Hence, in this double-blind randomized crossover study, the investigators aim to use caffeine, an adenosine antagonist as well as a commonly used psychostimulant by nearly 80% of the worldwide population, to examine whether blocking A2AR will enhance D2R-modulated reinforcement learning/no-go responses through modulating D2R signaling pharmacologically.

The long-term goal of this study is to further the understanding of molecular mechanisms related to A2AR-D2R heterodimers and the clinical potential of modulating A2AR-D2R interactions. Twelve young healthy non-smokers will enroll in this study. Each participant will undergo a caffeine and a placebo condition. In each condition, participants will first go through a 6-day ambulatory washout period where participants will be asked to abstain from caffeinated dietary, alcohol, and drugs, and stay in regular bed- and wakeup time. On day 7, a PET/fMRI scan will take place at noon, and a caffeine or placebo tablet will be administered orally 20 mins prior to the scan.

Simultaneous PET/fMRI will be used to examine the association between the neurochemical changes (i.e., D2/D3R availability as quantified by [11C]Raclopride) and the hemodynamic responses (i.e., task-related blood oxygen level-dependent fMRI activity) during reinforcement learning in the caffeine condition compared to placebo. It is hypothesized that enhanced D2/D3R availability mediates the facilitating effect of caffeine on reinforcement learning. Specifically, the investigators expect that caffeine will enhance fMRI responses in reward-related brain regions, and that the increased fMRI response will positively correlate with a change in D2/D3R availability.

Inclusion Criteria:

• Age ≥ 18 and ≤ 45.
• Habitual caffeine intake ≥ 100 mg and ≤ 450 mg daily.
• Non-smokers.
• Clinically healthy.
• Have normal vision or corrected to normal vision.

Exclusion Criteria:

• Pregnant or lactating women.
• Women using hormonal contraceptives.
• BMI < 18.5 or > 29.9
• Sleep disturbance or extreme chronotype.
• Urine test positive on one of the following substances: benzoylecgonine, morphine, d-Methamphetamine, d-Amphetamine, Benzodiazepines, Secobarbital, Methadone, Buprenorphine Glucuronide, Nortriptyline, MDMA, Oxycodone, PCP, Propoxyphene, and Cannabis/THC
• Diagnosis of depression, anxiety, psychosis, or neurologic disorders in the last 5 years.
• Heart or cardiovascular diseases.
• Diabetes or other metabolic diseases.
• Under chronic medications, for instance, painkiller and steroid.
• Allergy to lactose (main ingredient of blank control dose)
• Incapable to operate the tasks or comprehend the study information in English.

General MRI and PET safety exclusion criteria for all subjects:

• Metallic foreign bodies such as cardiac pacemakers, perfusion pumps, aneurysm clips, metallic tattoos anywhere on the body, tattoos near the eye.
• Pre-existing medical conditions including a likelihood of developing seizures or claustrophobic reactions
• Inability to lie flat on scanner bed for about 90 min as assessed by physical examination and medical history (e.g. arthritis)
• Recent exposure to radiation (i.e., PET from other research studies) that, when combined with this study, would be above the allowable limits
• Pregnancy or breastfeeding: A negative serum or urine pregnancy test is required on the day of the PET procedure
• Body weight of > 300 lbs (weight limit of the MRI scanner table)

Raw, subject-level data files will be submitted to NIMH Data Archive (NDA). These include [11C]Raclopride PET data as well as behavioral and fMRI data from Probabilistic Selection Tasks and Go/Nogo tasks. Other behavioral data e.g. Salience Attribution Tests, Karolinska Sleepiness Scale, and State-Trait Anxiety Inventory, as well as fMRI data e.g. Arterial Spin Labeling and resting state fMRI will also be included.

Relevant resources, such as code used for data processing and analyses, will be made publicly available through GitHub. The main readme.md file for the project will include instructions and parameter choices to execute the codes/scripts. Code documentation will include instructions on how to access data, the name and email address of a contact person for questions, and relevant references to publications. To ensure long-term accessibility, a copy of the GitHub code repository will be archived in Zenodo at the time of publication with a digital object identifier (DOI).

Data will be available at the end of the NIMH grant (anticipated 1/1/2027). Data and the code/software/tools used to develop the published or submitted dataset will be shared at the time of data submission or publication and maintained for no shorter than 5 years.