A Flexible Fluid Delivery System for Rodent Behavior Experiments

Syringe pump system. a, 3D model of the fully assembled syringe pump system. Controller printed circuit board (PCB), syringe, switches, and stepper motor are highlighted. The assembly process is documented with step-by-step instructions, including detailed photographs. b, Diagram of controller PCB. The three main sections of the board are highlighted: microcontroller, which implements the Harp protocol; motor driver and power, which provide the low-level logic to drive the stepper motor; and the I/O breakout, that affords users with input and output lines which can be used to control and monitor the function of the system, respectively. The modularity of the board design affords users the option to assemble a simpler version without the microcontroller block, for applications wherein low-level control is sufficient. This cheaper version is assembled using the same PCB schematic, with minimal changes to the board components, and can be found in an alternative bill of materials provided. See Materials and Methods for further details.

The current system affords three distinct levels of interface. a, Low-level hardware control expects the user to fully define the control logic for the stepper motor driver. This can be achieved by implementing such logic in a microcontroller (e.g., Arduino, right textbox) that defines the state of all input control pins to the stepper motor driver. It should be noted that this mode does not require the full PCB to be populated, since it does not rely on the Harp core protocol implementation. b, Trigger-based control allows the user to use an external trigger event (via transistor–transistor logic, TTL) to playback a pre-defined delivery protocol. The default protocol values (e.g., volume and flow rate) can be modified using a provided graphical-user interface, or Bonsai. c, Software control allows the user to fully parameterize, and trigger, protocols from a computer host running Bonsai without the need for any external hardware triggers. Sub-millisecond synchronization can be achieved via Harp protocol, or by a configurable digital output event.

Tracking small single-bolus events. a, Schematic of the computer vision algorithm for measuring small microliter range volumes. From top to bottom: cropped, thresholded, and segmented the area of the capillary filled with liquid. Area was taken as a proxy of delivered volume (see Materials and Methods for further details). b, Example trace of the measured area in a, as a function of time during one of the experiments. Red vertical dashed lines represent liquid delivery events.

Single-bolus protocol calibration. a, Time course of displaced area aligned on protocol onset (t = 0) for four distinct theoretically expected delivered volumes. Thin and thick lines correspond to single trials, and averages for a given expected volume, respectively. Shaded area depicts s.t.d. b, c, Total displaced area for the four protocols. Each point shows mean ± s.t.d. across 30 replicates, per volume, in two different pumps (Pump A and Pump B, green and red, respectively) and using two different glass syringe sizes (5 and 10 ml, b and c, respectively). d, e, Coefficient of variation (s.t.d./mean) calculated from the data shown in b and c, respectively.

Syringe pump affords dynamic control over the flow rate. a, Time course of delivered volume (displaced pixel area) aligned on protocol onset (t = 0) for different inter-protocol-interval values (IPI). Thin and thick lines represent single trials, and averages, for a given IPI, respectively. b, Estimated flow rate (pixels s⁻¹) for all tested IPI (mean ± s.t.d., n = 6 trials for each IPI).

Large volume protocol calibration. a, Delivered volume (mean±s.t.d.) across 20 performed delivery protocols for four distinct large volume amounts, in two independent pump systems (Pump A and Pump B, green and red, respectively). “Predicted volume” was calculated by weighting the delivered liquid and assuming a water density of 1 g/ml. “Expected volume” was calculated using the theoretically displaced syringe volume. R² and slope (β) resulting from the linear fit to the data are shown in the top-left corner. b, Coefficient of variation (s.t.d./mean) calculated from the data shown in a.

Comparison of calibration stability between the gravity-based solution and the presented pump system. a, For each system, the calibration (see Materials and Methods) consisted of measurements of delivered liquid as a function of the number of pre-programmed delivery protocols, or valve opening time, for the Pump (left) and Valve (right) systems, respectively. Individual fits correspond to a single-day calibration protocol. For each system, we tested two devices independently (distinct colors). b, Predicted delivery volume for two arbitrary volumes (10 and 30 μl, top and bottom, respectively) based on the calibration linear fits derived from a. For each day, the fit calculated from day 0 was used as to infer that value would have been delivered had the calibration curve remained the same.

Rats quickly reverse choice preference on delivered liquid reward amount. a, Example session. Animals alternate between the two available choices (“X” and “notX,” respectively) in a way that appears consistent with the highest-value choice (shaded area). Thick dark line depicts the probability of choosing “X” in a rolling window of 15 trials. b, Probability of choosing reward giving nose-port X as a function of the possible reward amount combinations given by each nose-port. Values in the heatmap axes are the number of protocols given on each reward delivery, by each reward nose-port (X and notX). c, Probability of choosing the highest rewarded side over the trials following block transition. Lines correspond to the absolute value of the difference between rewards at each nose-port in logarithmic units (base 2). The highest the difference between the two rewards available on a particular block, the sooner animals converge to the highest rewarded nose-port, suggesting the need for fewer reward samples (trials) as the difference between rewards increases.