DIY-NAMIC behavior: A high-throughput method to measure complex phenotypes in the homecage

Complex behavioral assessment is becoming increasingly necessary in order to comprehensively assess in vivo manipulations in rodent models. Using operant behavioral paradigms provides rich data sets allowing for the careful analysis of behavioral phenotypes. However, one major limitation in these studies is the expense and work-load that are required using traditional methods. The equipment for commercial operant boxes can be prohibitively expensive, and the daily experimenter effort and mouse costs required for these studies is extensive. Rodents are generally trained on task-specific paradigms for months, tested every day for 5-7 days per week. Additionally, appetitive paradigms usually require food restriction and are also commonly run in the non-active light phase of the rodent circadian rhythm. These limitations make operant behavioral testing especially difficult during adolescence, a time period of interest with regards to the development of adult-like phenotypes and a high-risk period for the development of neuropsychiatric disorders, including those which involve impulsive behavior. In order to address these issues, we developed an automated, inexpensive, open-source method which allows the implementation of most standard operant paradigms in the homecage of rodents in shorter time frames without food restriction, and with much less experimenter effort. All construction and code for the DIY Nautiyal Automated Modular Instrumental Conditioning (DIY-NAMIC) system are open source. We demonstrate their utility here by measuring impulsive behavior in a pharmacology experiment, as well as in adolescent mice. Significance statement Rigorous behavioral assessment is critical to understand the neural basis of neuropsychiatric disorders using animal models. Operant behavioral paradigms provide the ability to measure complex phenotypes, however, traditional methods generally require time-consuming daily training for many weeks. We designed, built, and tested an open-source automated homecage system for appetitive instrumental conditioning that enables testing in shorter timeframes with reduced experimenter effort.

In order to measure impulsivity and a range of related behavioral parameters, we sought a more time-and cost-effective method for fine-grain measurement of complex behavior in rodents. In recent years, there have been a number of automated open-source developments for rodent testing relevant to our work, including home-cage feeding and fluid consumption [34][35][36]. 'do-it-yourself' (DIY) operant behavioral testing apparatus have been developed for use with standard session-based testing methods which provide relatively inexpensive and customizable apparatus making behavioral testing more accessible [37]. Additionally, there have been modifications made to traditional commercial equipment, and also to standard lengthy paradigms making shorter training times possible [38,39]. What was still lacking, however, was a method that combined these advantages to allow for both homecage and continuous assessment of operant behavior, with liquid reward in an affordable open-source system.
Our goal was to design, build, validate, and disseminate a novel method which allows for continuous quantitative and robust operant behavioral testing. We developed the DIY Nautiyal Arduino Modular Instrumental Conditioning (DIY-NAMIC) system as an inexpensive, opensource method which allows mice to perform operant trials in their homecage to receive their daily consumption of water, in self-initiated trials. This promotes rapid acquisition of taskspecific learning in behavioral paradigms given the high number of daily trials that mice perform each day. With continuous testing and automatic data-logging the DIY-NAMIC system enables high throughput data collection 24h/day, 7d/week. Using the DIY-NAMIC system, we were able to measure impulsive action as well as a number of other behavioral parameters in under three weeks, with only twice weekly (less than 1h long) experimenter effort. We show that mice rapidly acquire Pavlovian and stimulus-response associations, and basic cue-based discrimination learning within a week. Furthermore, we show that effects of pharmacological manipulations can be measured with the DIY-NAMIC system, and demonstrate that activating the serotonin 1B receptor (5-HT 1B R) reduces impulsive action. Finally, our studies illustrate the ability to test impulsivity in adolescent mice, a period of development which is difficult to test in mice using standard paradigms. Overall, the studies presented here validate the use of the DIY-NAMIC system for assessing multi-dimensional behavior including impulsivity. Furthermore, we present novel data showing serotonergic modulation of the neural basis of impulsive behavior, and by characterizing adolescent impulsivity in mice.

DIY-NAMIC System Overview
The DIY-NAMIC system was built for compatibility with most facilities' standard rodent homecages in high-density ventilated racks by replacing one cage wall with a modular system that delivers stimuli and rewards, and detects and records behavioral responses. It allows for the implementation of standard Pavlovian and operant-based tasks though with less effort, time, and cost. The DIY-NAMIC system is controlled by an Arduino microprocessor board and easily sourced electronics allowing for an inexpensive and customizable apparatus. The initial design used in these studies includes three solenoid-delivered liquid dispensing noseports, each with an infrared head-entry detector and LED cue light. Mice do not undergo food or water restriction and the DIY-NAMIC system permits behavioral testing during the active phase of their circadian rhythm and eliminates daily, potentially stressful, handling. The system is designed for liquid reward delivery enabling fine control of volume (versus pellet delivery), and can thus also be used with water or palatable liquid rewards. The receptacles are shallow, allowing reward receipt with large head stages, such as those for miniature head-mounted microscopic imaging. Analysis scripts, written in Python, are available for use and modification as needed. Given their modular construction and open source development, the hardware for the DIY-NAMIC system is customizable and inexpensive to build.

DIY-NAMIC Build Instructions
All materials required to build DIY-NAMIC boxes are easily sourced from common websites with the total cost of all parts required for each box at less than $300 (Table 1). All build instructions, detailed diagrams, and design files are located online (https://github.com/DIY-NAMICsystem). Standard mouse ventilated home cages (5.5" x 14" x 5", Techniplast) were modified by cutting one short wall off with a rotary tool, and U channels were affixed with Loctite 401 inside the cut edge. A plexi-glass wall was laser cut to fit the dimensions of the cage, and slotted into the U channels. The wire and ventilated cage tops fit as usual. Three nose ports per cage were 3D printed and attached to the plexi-glass wall with flat head screws. Photo interrupters were cut in half to separate the infrared LED and phototransistor and each part was inserted into casing on either side of the designed noseport. LEDs with current-limiting resistors were secured into the top of the back wall of each nose port. Large gauge hypodermic needles (blunted by removing the beveled sharp tip) were inserted into the lower hole on the back of each nose port. Luer locks attached the needle to tubing connected to a solenoid valve. The three solenoids (one per noseport) were connected to a 3-way manifold which was fed water through tubing from a 60ml syringe.
An Arduino UNO R3 board was connected via double stacking header pins to an Arduino "shield" from OpenMaze.org (OM1 board), which allows easy connection of components such as LEDs and IR beam break detectors, and contains basic circuitry to allow the Arduino to drive higher-voltage components (e.g. an "H-bridge" for solenoid valves). An integrated circuit socket for an H-bridge and DC Barrel Jack were soldered to the board to provide power for these components. The LED, solenoid, and IR detectors were wired to the OM board with jumper cables and Dupont connectors as shown in the circuit diagram in the build instructions. A wall power supply was connected to the board through the DC barrel jack connector. For multiple cages running in parallel, Arduinos were connected via 10-USB hubs to a PC to allow for continuous data logging of 16 cages directly to a PC using Processing software.

DIY-NAMIC Data Collection and Processing
Behavioral responses, as well as stimuli and reward presentations are logged with event and timecode stamps which are written directly into text files through Processing scripts. Following data collection, initial analysis is performed off-line using Python scripts in order to concatenate data log files, and extract relevant metrics including total number of rewards, and total number of nosepokes to each port, and nosepokes to each port during the ITI, reward presentation, cue presentation, and delay period. Premature pokes were considered pokes to one of the two side ports during the delay period. Time bounds are input to allow data extraction in desired time bins (e.g., minutes, hours, dark cycle, or days). All data collection and processing scripts are available online (https://github.com/DIY-NAMICsystem).

Mice
Mice were bred in the Moore Hall vivarium at Dartmouth College. Mice were weaned at postnatal day (PN) 21 into cages of 2-4 same sex littermates and maintained on ad libitum chow and water in a 12:12 light-dark cycle. For the initial characterization of behavior shown, male and female mice (N=9) were tested in DIY-NAMIC boxes beginning at 14 weeks of age. For the study of impulsivity in adolescents, male (N=9) and female (N=9) mice were placed individually in DIY-NAMIC boxes at 45-47 days postnatal (Adolescents, N=9) or 14-15 weeks postnatal (Adults, N=9). The pharmacology study was performed on mice aged 11-20 weeks old (N=18). All mice were placed in DIY-NAMIC boxes individually, with environmental enrichment including a plastic igloo and nesting material with ad libitum chow. All mice were on a mixed C57/Bl6J and 129Sv/Ev background. All procedures were approved by the Dartmouth College Institutional Animal Care and Use Committee.

Behavioral Training
Mice were singly housed and initially habituated to the DIY-NAMIC boxes and to retrieving water from the center noseport delivered by a solenoid for 3 days. During this reward retrieval training ("P1" ; Fig 2A), the cue light in the center port was illuminated and mice received 10ul of water upon poking into the center port. After water retrieval, a variable inter-trial interval (ITI), with an average of 45s preceded the next light illumination. The ITI remained at 45s average for all paradigms. Any mice not receiving more than 100 rewards over 24h during center port training was removed from the DIY-NAMIC cage (only 2 mice out of 46 did not meet this criteria). Subsequently, mice were trained on a continuous reinforcement schedule ("P2"; Fig  2D), for 3 days to nose poke in one of the two side nose ports when illuminated with a cue light. Pokes were rewarded in the center port. Both ports were illuminated with cue lights indefinitely on each trial, and a poke to either port was rewarded. A trial initiation requirement was then added ("P3"; Fig 2G), and each trial began following a nosepoke to the center port blinking (1 Hz). Next, a basic cue discrimination requirement was added ("P4"; Fig 2I) in which only one of the two sides were illuminated, and only pokes to the illuminated port were reward. For 3 days, only correct nosepokes produced an outcome (reward) and incorrect pokes (pokes to the non-illuminated port) had no consequence. Subsequently, incorrect pokes resulted in trial termination and ITI onset, and trials were capped at 5s (no response before 5s resulted in ITI onset and the trial was considered an omission. After 2 days, the trial response duration was shorted to 1.5s. Finally, a modified 2-choice serial reaction time test paradigm (Fig 2K) was tested as a measure of impulsive action, by introducing a delay period between trial initiation and the cue light onset. The time delay was increased over 3 lengths: 3s, 6s, and 9s -with each presented for 3 days. Nose pokes during the delay were recorded but had no influence on the outcome of the trial (i.e. there was no timeout/punishment period or cue). Only a poke during the illuminated cue was necessary for a rewarded trial. All programs were switched during the light cycle, roughly 4-6h after light onset. Arduino programs for all paradigms are provided online including schematics illustrating trial structure and detailed input/output descriptions (https://github.com/DIY-NAMICsystem).

Drug administration
Following training as described above, and immediately following 3 days of exposure to the 9s delay in the test of impulsive action, mice were injected with drug or saline control. A selective 5-HT 1B agonist, CP 94253 hydrochloride (Tocris Bioscience Cat. No. 1317) was given at a high dose -10 mg/kg, dissolved in 0.9% sterile saline and injected at 10ml/kg i.p [40]. Injections were given within 30min of initiation of the dark cycle. Mice were randomly assigned to drug or vehicle conditions, and then 2-3 days after receiving their first injection, mice received the alternate condition. Data from the 12 hours (their dark phase) after receiving injections were analyzed.

Statistical Analysis
Statistical testing was performed using analysis of variance (ANOVA), with post hoc Fisher's least significant difference (LSD) in StatView (SAS Software, Cary, NC) or SPSS (IBM, Armonk, NY) for three-way ANOVAs. For the initial behavioral characterization, a repeated measures ANOVA was used to assess the change in behavior over days in the measures of number of rewards, number of ITI responses, and number of trials initiated per day. Post hoc Fisher's LSD was used to assess differences between days. For the adolescent study, a three-way repeated measures ANOVA was used to test the effect of sex and age on premature responding. There were no significant effects of sex, so the groups were combined for the remaining analysis. Measures were summed over each day and then averaged across 3 days for each delay, except for proportion of correct trials which was averaged across each hour of the day, and then averaged across 3 days for each delay. Two-way repeated measures ANOVAs, with post hoc Fisher's LSD tests, were used to assess the effect of age on four behavioral measures over the 3 delay lengths [age (adol, adult) x delay length (3s, 6s, 9s)]: premature responding (nosepokes during the delay window), number of trials initiated, number of omitted trials, and proportion of correct trials. For the pharmacology study, premature responses per trial was calculated by dividing the number of nosepokes during the delay period by the number of self-initiated trials for each hour, and a repeated measures ANOVA [condition (drug, saline) x time (1-12h)] was used to assess significance. All other measures were summed (number of delay responses, ITI responses) or averaged (proportion of correct responses and omissions), over 6h bins following injection (1-6h and 7-12h following injection) to analyze the effect of drug versus saline, based on drug half-life estimations from previous microdialysis studies following CP 94253 drug administration [40]. Then, repeated measures ANOVAs [condition (drug, saline) x time (1-6h, 7-12h)] were used to assess the effects of CP 94253 over the time, with post hoc Fisher's LSD.

Results
We designed, implemented, and tested the DIY-NAMIC system, and show that it is a low-effort, low-cost, and high-throughput method to measure complex behavior in mice (Fig 1). Slotted into a standard rodent home cage compatible with high-density racked cages, the DIY-NAMIC system allowed mice to rapidly learn complex operant behavioral tasks. Mice learn to nose poke at a cue-lit reward port to receive a liquid reward on a continuous reinforcement schedule (Fig 2A), receiving an average 172±13 rewards per day over the first three days of testing. Compared to the average 40-60 rewards/day in standard operant behavioral designs, this is a large increase in the number of trials/day that are run, therefore greatly speeding task acquisition. The majority of responses made were during the dark phase of their diurnal cycle; mice received only 10.6% of their rewards during the light phase (Fig 2B). This allows for more naturalistic testing of nocturnal rodents compared to many standard protocols which involve behavioral testing during the mouse dormant phase (daytime) for facilities that don't have reverse light cycles. Additionally, intertrial interval (ITI) responding decreased by the second day ( Fig 2C; F 2 Mice were subsequently trained to make an operant response (nosepoke) to an illuminated response-port (Fig 2D), and increased their earned rewards by the second day ( Fig  2E; F 2 (Fig 2G), and increased to stable performance by the second day (F 3,24 =5.7, p<0.01; D1 vs all other days: p<0.05) self-initiating an average of 191±10 trials per day ( Fig  2H). Following trial self-initiation training, mice learned to respond correctly on a discriminative cue paradigm, during which a response only in the illuminated port resulted in a reward (Fig 2I). Mice performed correctly by responding with a nosepoke to the illuminated port on 84.7% of the 211.3±12.2 total trials, with incorrect responses on 13.8% of trials, and omissions on only 1.4% of trials (Fig 2J). Next, we assessed the capability of the DIY-NAMIC system to measure impulsive behavior using a modified 2-choice serial reaction time task (Fig 2K). Delays of varying lengths were introduced between trial initiation and the response cue in order to measure premature responding. The number of premature responses varied as a function of delay length (F 2,16 =3.5, p<0.05; Fig 2L). Specifically, during the dark period (when mice initiated the majority of trials), there were more premature responses per trial during the 6 and 9 second delay, compared to the 3 second delay (3 vs 6: p<0.05; 3 vs 9: p<0.01; 6 vs 9: p>0.05). Over days within a delay, the accuracy did not change (F 2,16 <1.53, p>0.05), however, omissions decreased across days during the 3 second delay (F 2,16 =14.3, p<0.001; Fig 2M).
We next assessed impulsivity during adolescence, given that the adolescent period in mice is generally shorter than the time needed for training on many traditional operant paradigms. While our goal was to measure impulsivity, it was important to also obtain additional measures of attention, motivation, and performance. Compared to adults, adolescent mice showed increased impulsive behavior as measured by premature responding (Fig 3A; F 1,16 =8.54, p<0.01) with more premature responding in adolescents as the delay got longer (F 2, 32 =24.9, p<0.001). Specifically, adolescents had significantly higher levels of responding during the 9 second delay period (p=0.05 for 3 and 6 sec; p<0.05 for 9 sec). This suggests that premature responding is a readout of the adolescents' reduced ability to inhibit responding since the number of responses increases as the difficulty/delay increases, rather than a reflection of their observed general hyperactivity or increased motivation. Importantly, adolescents did initiate more trials (Fig 3B; F 1,16 =5.8, p<0.05), however this did not vary with the change in delay/difficulty of the paradigm (F 2,32 =1.4, p>0.05). We also used the number of omitted trials as a readout of attention. Adolescent mice showed decreased attention as measured by increased number of omission trials ( Fig 3C, F 1,16 =7.1, p<0.05). Again, there was no effect of delay length on omissions suggesting that the decreased attention was unlikely the cause of the increased impulsivity (F 2,32 =1.2, p>0.05). Importantly, adolescents did not show differences in the proportion of correct trials ( Fig  3D, F 1,16 =0.1, p>0.05) indicating that their performance on the task was normal and there was no effect of age on accuracy or overall performance on the paradigm. Overall, the results suggest that adolescent mice display increased impulsive action as measured by premature responding. Furthermore, these data demonstrate the ability to test complex self-initiated operant behaviors during limited time frames, such as adolescence, in the DIY-NAMIC system.
Finally, in a pharmacology experiment, we tested the effect of manipulating serotonin signaling on impulsive action using DIY-NAMIC boxes. We focused on the effect of activating the 5-HT 1B receptor by administering the receptor-specific agonist, CP 94253, and measuring behavior using a modified 2-choice serial reaction time task in the DIY-NAMIC cages. This method enabled more extensive measurement of behavior over long time periods (throughout 12h of the dark cycle) following drug administration. Compared to operant behavior which traditionally measures behavior at single point in time, the DIY-NAMIC system allowed investigations into the timecourse of drug effects on behavior (Fig 4A, main effect of time: F 11,374 =3.3, p<0.001). Administration of the agonist resulted in decreased impulsivity, measured by reduced premature responding, which varied over time as the drug effects dissipated (Fig 4B; interaction of drug x time: F 1,34 =4.81, p<0.05). Specifically there was decreased responding during the delay period in the 6h following injection of the 5-HT 1B receptor agonist (p<0.05), and not during the subsequent 6h of the dark phase (p>0.05). Importantly, the agonist also decreased the total number of trials initiated in the first 6h after drug administration, suggesting a decrease in general motivation (Fig 4C, interaction: F 1,34 =10.6, p<0.01; drug vs vehicle for 1-6h: p<0.05). While this is a relevant factor in interpreting the premature responding, the correction for premature responses per trial (Fig 4A), suggests that the drug effects on impulsive responding is not solely driven by decreases in motivation. Additionally responding during the ITI can be driven by both impulsive and general hyperactive behavior; these responses were slightly, but not significantly decreased for the first half of the dark cycle following drug treatment (Fig 4D, interaction drug x time: F 1,34 =3.1, p=0.09). Finally, there was also a significant effect of the agonist on the proportion of omitted trials (Fig 4E; F 1,34 =6.3, p<0.05), and this was only significant for the first 6h following drug administration (interaction of time x drug: F 1,34 =4.5, p<0.05). There was no effect of the agonist on the proportion of correct trials of those attempted (Fig 4F; F 1,34 =2.0, p>0.05), indicating the drug effects were specific to premature responding, and likely not due to effects on performance or attention. These data show the utility of DIY-NAMIC boxes in assessing pharmacological manipulations, particularly in allowing for timecourse data. Additionally, these data show a novel effect of serotonin pharmacology on impulsivity.

Discussion
Given the recent push to explore more robust and multi-dimensional behavioral readouts of animal models for neuropsychiatric disorders, the development of high-throughput and automated behavioral screens are critical [1,41]. Operant behavioral paradigms provide rich behavioral datasets, however they are labor intensive and lengthy (daily training for weeks/months), and generally require food deprivation and costly equipment. To address these limitations, we developed an Arduino-based system as a homecage, low-touch, fully automated operant chamber that allows mice to self-initiate trials for their daily water intake. Data is logged continuously to a PC, and experimenter effort is only required to monitor/weigh animals to assess welfare, and upload new programs to Arduinos when switching behavioral paradigms (though even this could also be automated in future implementations). We have shown that mice learn basic Pavlovian and operant behaviors over a few days in DIY-NAMIC cages, performing hundreds of trials in a day in their dark/active phase. Additionally we have validated its use in measuring effects of drugs on behavior, allowing for monitoring over long time-frames. Finally, we show that we can measure impulsive behavior in adolescent mice, and show increased impulsivity compared to adults.
The difficulty of training mice on complex operant behavioral paradigms within a few weeks using standard procedures and equipment limits their utility in examining adolescent behavior in mice given its short timeframe [16,42]. We have found this to be a critical limiting factor in understanding the adolescent development of the neural and behavioral basis of impulsivity, an important sensitive period for the etiology and pathogenesis of neuropsychiatric disorders which include impulsive behavior as a core phenotype [43,44]. Our studies reported here measure behavior in mid-to late-adolescence, though the DIY-NAMIC system easily allows future studies to characterize impulsivity throughout various stages of adolescence including early adolescence (~30-40 days).
There have been past efforts to study operant behavioral measures in adolescent mice. Some of these studies have taken the strategy of eliminating mice that don't learn quickly enough to progress to increasingly complex behavioral paradigms, leading to potentially biased samples [45]. There has also been some progress in the development of new training paradigms that reduce training time to allow testing in adolescent mice [38,46], however they still require food deprivation which could be detrimental to growth and the maturation of feeding circuits during the adolescence. Inexpensive Arduino-based operant testing apparatus have also been developed, such as ROBucket [37], though these have also been used with the daily experimenter-initiated daily sessions. Finally, others have used modified commercially available designs such as the CombiCage which allow free-access to the standard costly operant conditioning chambers via a tube connected to a homecage to measure adolescent impulsive action [39]. Overall, the DIY-NAMIC system provides an improved method to measure impulsivity in adolescent mice by combining the benefits of some preceding technology and allowing testing in a number of behavioral domains with minimal experimenter effort and highthroughput protocols. We were able to use the rich data set collected to tease impulsivity apart from other behavioral parameters that may also be different during adolescence and could otherwise obscure or contaminate measures of impulsive behavior. For example, previous work has shown hyperactivity, increased motivation, and decreased attention in adolescent mice [46]. We were able to increase the sensitivity of the task to detect impulsivity by increasing the delay length [47]. Our measure of impulsivity increased as delay length increased, while measures of hyperactivity, motivation, and attention did not.
Automated home cage phenotyping also provides the opportunity to measure behavior in a more reproducible manner because it limits experimenter-introduced variability [48]. A system that is compatible with high-density racked cages and does not include video-based phenotyping (because of the long time-frames) provides the most usable and reliable method. Additionally, homecage behavioral testing allows for behavioral analysis across the circadian cycle, an important variable for the interpretation of studies assessing impulsivity, that is often overlooked [49]. Conventional session-based, non-homecage behavioral testing generally occurs during the light phase of the circadian rhythm, likely causing partial entrainment to feeding occurring in the light cycle, and resulting in dysregulation of homeostatic systems [50]. Furthermore, it is difficult to test the effect of circadian rhythms using traditional session-based operant paradigms which is critical because many manipulations, including drug administration, have interactions with time-of-day.
The DIY-NAMIC system also provides a data-rich, low-effort approach to psychoactive drug screening. We show here that the effects of acute administration of drugs can be assessed over prolonged periods of time. This is an important improvement compared to many "singleuse" behavioral tests which can only reliably be performed once, requiring higher numbers of animals in order to be able to look at the temporal profile of drugs. Additionally, the DIY-NAMIC system provides an opportunity to assess longer-term effects of acute administration, as well as effects of chronic drug administration, with lower effort and animal numbers than traditional testing. Finally, the shortened training time required allows for the preclinical testing of drugs in adolescence, which is important given potential differential actions of drugs over the lifecourse.
Although the DIY-NAMIC system provides a number of significant advantages over standard commercially available operant boxes, one downside currently is the need to singlehouse mice. Although the daily stressor of handling is eliminated, singly-housed mice may also result in effects on baseline stress levels and/or behavioral readouts [51][52][53]. The DIY-NAMIC system is compatible with standard homecage enrichment, and we include nesting material and an igloo hut in all cages. The goal of future iterations of this method will involve the ability to group house mice with an RFID chip implanted subcutaneously, and a reader in an access port to the receptacles. This could allow standard group housing by litter or in pairs. Additionally, we use the DIY-NAMIC system here to measure impulsive action, but many other paradigms can be employed using this method. Our next steps include the implementation of random ratio and progressive ratio schedules of responding, and other types of impulsivity, such as delayed discounting. Finally, given the Arduino platform and the modular build of the DIY-NAMIC system, additional stimuli and cues (e.g. vibration pad for tactile stimulus) are possible for customization for measurement of different phenotypes.
In conclusion, we have developed the DIY-NAMIC system as a low-touch, low-cost homecage operant behavioral testing system with an opensource platform. All materials are commercially available, and build instructions are documented online in detail, including tips for researchers with minimal electrical and material engineering skills. Software for all of the programs and analysis used for experiments in this manuscript are also available online, including Arduino programs for running behavioral paradigms, Processing scripts for logging data, and Python scripts for data analysis and plotting. The DIY-NAMIC system greatly increases the ability to measure complex and robust behavioral measures in a high-throughput manner, enabling more productive and reproducible basic research in animal models for neuropsychiatric disorders.