Dopamine axon activity in the ventral, dorsomedial, and dorsolateral striatum was recorded, while mice performed a perceptual and value-based decision-making task. In one experiment, thirsty mice performed a perceptual decision-making task using mixtures of odor A and B (100/0, 90/10, 65/35, 35/65, 10/90, 0/100), in which identity of a dominant odor determined an available water port, and odor C which signaled no outcome. A fixed amount of water was always delivered with a correct choice. In the next experiment, mice performed a perceptual and value-based decision-making task using mixtures of odor A and B (100/0, 65/35, 35/65, 0/100), in which identity of a dominant odor determined an available water port, with probabilistic water reward. In this task, a fixed amount of water was delivered in block 1, and then in block 2, one water port delivered big or medium size of water in a pseudo-random order, and another water port delivered medium or small size of water in a pseudo-random order. In both tasks, odor-water port (left or right) rule was held constant throughout training and recording in each animal.

Fiber fluorometry (photometry) was performed as previously reported (Menegas et al., 2018) with a few modification. The optic fiber (400 µm diameter, Doric Lenses) allows chronic, stable, minimally disruptive access to deep brain regions and interfaces with a flexible patch cord (Doric Lenses, Canada) on the skull surface to simultaneously deliver excitation light (473 nm, Laserglow Technologies, Canada; 561 nm, Opto Engine LLC, UT) and collect GCaMP and tdTomato fluorescence emissions. Activity-dependent fluorescence emitted by cells in the vicinity of the implanted fiber’s tip was spectrally separated from the excitation light using a dichroic, passed through a single band filter, and focused onto a photodetector connected to a current preamplifier (SR570, Stanford Research Systems, CA). During recording, optic fibers were connected to a magnetic patch cable (Doric Lesnses, MFP_400/430) which delivered excitation light (473 nm and 561 nm) and collected all emitted light. The emitted light was subsequently filtered using a 493/574 nm beam-splitter (Semrock, NY) followed by a 500 ± 20 nm (Chroma, VT) and 661 ± 20 nm (Semrock, NY) bandpass filters and collected by a photodetector (FDS10x10 silicone photodiode, Thorlabs, NJ) connected to a current preamplifier (SR570, Stanford Research Systems, CA). This preamplifier output a voltage signal which was collected by a NIDAQ board (National Instruments, TX) and Labview software (National Instruments, TX).

To synchronize behavioral events and fluorometry signals, TTL signals were sent every 10 s from a computer that was used to control and record task events using Labview, to a NIDAQ board that collects fluorometry voltage signals. GCaMP and tdTom signals were collected as voltage measurements from current preamplifiers. Green and red signals were cleaned by removing 60Hz noise with bandstop FIR filter 58-62Hz and smoothing with moving average of signals in 50ms. The global change within a session was normalized using a moving median of 100s. Then, the correlation between green and red signals during ITI was examined by linear regression. If the correlation is significant (p<0.05), fitted tdTom signals were subtracted from green signals.

Mice were first trained only with pure odors and with the same amounts of water reward (~6 ul). After mice achieved greater than 90% accuracy, mice received a surgery for viral injection and fiber implantation. Following a 1-week recovery period, mice received re-training and then, mixtures of odor A and B (100/0, 90/10, 65/35, 35/65, 10/90, 0/100) were gradually introduced. After the accuracy of all the mixture odors achieved more than 50%, neuronal recording with fiber fluorometry was performed for 5 sessions. Subsequently, a task with different amounts of water was introduced. Mixtures of odor A and B (100/0, 65/35, 35/65, 0/100) but no odor C were used in this task. Each recording session started with 88-120 trials with an equal amount of water (~6 ul, the standard amount) in the first block to calibrate any potential bias on the day. In the second block, different amounts of reward were delivered in each water port. In order to make the water amounts unpredictable, one water port delivered big or medium size of water (2.2 and 0.8 times of the standard, ~13.2 and 4.8 ml, BIG side) in a pseudo-random order, and another water port delivered medium or small size of water (0.8 and 0.2 times of the standard, ~4.8 and 1.2 ml, SMALL side) in a pseudo-random order. Block 2 continued for 200 trials or until the end of recording sessions, whichever came earlier. A mouse performed 134.3 ± 3.4 (mean ± SEM) trials in block 2. The water condition (BIG or SMALL) was assigned to a left or right water port in a pseudo-random order across sessions. Recording was conducted for 40 min every other day to avoid potential bleaching. On days with no recording, animals were trained with pure odors A and B with the standard amount of water.