Association for Behavior Analysis International

Computer technology, artificial intelligence, and machine-learning are not new tools to behavior analysis, especially in the experimental realm. However, there is incredible untapped potential for these tools to aid both basic and applied scientists in their understanding and treatment of behavior. One primary area of interest to incorporate computers and artificial-intelligence is in data-recording. In the basic realm, programming technology to provide continuous streams of data allows for a finer-grain analysis of the impact of different schedules of reinforcement on animal and human behavior. In the applied realm, relinquishing data recording of problem behavior to a computer or artificial intelligence allows for direct-support staff to focus more on treatment implementation. This symposium describes three studies that either create or extend computer programming and artificial-intelligence as tools to record animal or human behavior. The first two studies describe the creation and application of computer programming to investigate human relational behavior during a transposition task and interindividual spatial behavior with rats. The second study attempts to reverse engineer existing technology (anomaly-detection cameras) and extend it to the detection and recording of problem behavior in children with autism.

Audience members who are interested in artificial-intelligence applications to the science of behavior will benefit from a basic understanding of computer-related terminology. Audience members with a basic understanding of transposition tasks and methods to study relational behavior will enjoy this talk.

Operant studies of interactions between individuals have focused on evaluating the effects of reinforcement schedules based on discrete responses, with little attention to the impacts on interindividual spatial dynamics or the use of spatial responses relevant to establishing interindividual behavior. Using real time recording of organisms’ movement, our lab has reported that individual spatial dynamics are sensitive to reinforcement schedules based on spatial features, such as travel distance, which suggests that interindividual spatial dynamics may also be affected by reinforcement schedules based on spatial features, such as distance between subjects. The purpose of this experiment was to determine if a schedule of reinforcement, based on distance between conspecifics, would affect interindividual spatial behavior. Using a real-time sensing system, water reinforcement was delivered under a Fixed Distance (FD) 15 cm between subjects schedule, for six Wistar water-deprived rats. The results indicated an increase in approach responses between subjects during the FD15cm schedule, whereas it decreased when food was absent or presented independently of distance. We discuss our findings in terms of their contributions to advancing the study of inter-individual behavior, the utilization of new technologies for continuous sensing and response recording, and the translational implications for the field of social behavior.

Behavior frequency data is collected for individuals with autism to monitor progress and inform decision making by behavior analysts. This data impacts important life decisions, such as restrictive procedures fading, psychotropic medication prescribing, and residential placement (Vollmer et al. 2008). Studies have demonstrated exorbitant training and monitoring hours are required from a behavior analyst to ensure this data collected by direct care staff is accurate (Mozingo et al. 2006; Reis et al. 2013). If an artificial intelligence (AI) camera software can be programmed to detect the rate of target behaviors from individuals served, this will likely increase the accuracy of behavior data and inform better treatment, while also reducing the workload of direct care staff. This project will extend Lesser, Luczynski, & Hood’s 2019 study which used an AI camera to detect sleep disturbances in learners with autism. In that study, all movements exhibited by the participants were recorded by the camera. This extension aims to program the camera to discriminate successive approximations of analog topographies of gross motor movements such that the camera will calculate instances of head directed self-injury but will ignore other topographically similar movements, such as hand waving.