The Big Picture

The criminal justice system in the U.S. has long sought methods to assess and manage the risk of reoffense among offenders. Traditional risk assessments often rely on subjective interviews, self-reporting, or criminal records, which can be time-consuming and susceptible to biases. The NeuroCognitive Risk Assessment (NCRA) aims to address these issues by using a tablet-based, gamified suite of neurocognitive tests to predict recidivism. This tool measures decision-making traits such as impulsivity, empathy, and aggression, which are associated with criminal behavior. The NCRA provides a scalable, objective, and potentially less biased alternative to traditional methods.
What We Did

The study introduced the first mobile, self-scoring risk assessment software, which uses neurocognitive tests to predict reoffense. The assessment, administered on tablets, was tested on 730 probationers in Houston, TX, from 2017 to 2019. By employing machine learning algorithms, the researchers evaluated the predictive validity of the NCRA, comparing it with traditional risk assessments. The key features assessed included attentiveness, aggression, risk-seeking behavior, empathy, future planning, emotional processing, and impulse control.

1

Predictive Validity:

The NCRA achieved a recidivism prediction accuracy with an AUC (area under the curve) value of 0.70, comparable to traditional risk assessment tools. This suggests that the NCRA can effectively classify individuals as likely to recidivate or not, similar to other established methods.

2

Scalability and Efficiency:

The software is self-administered and requires no specialized training to operate, making it easier to deploy on a large scale. It allows for group testing and significantly reduces administrative time compared to traditional one-on-one interviews.

3

Reduction of Bias:

By focusing solely on neurocognitive traits and excluding factors such as race, socioeconomic status, and previous arrests, the NCRA aims to reduce biases that are often present in traditional risk assessments.

The use of objective, data-driven neurocognitive assessments can enhance the fairness and accuracy of risk evaluations in the criminal justice system. The NCRA has the potential to streamline processes, reduce costs, and provide more individualized sentencing and rehabilitation strategies, ultimately contributing to more effective management of recidivism and public safety.

1

Expand Testing and Deployment:

Broaden the use of NCRA across various points in the criminal justice system, such as pre-trial assessments, sentencing, probation, and parole.

2

Improve Machine Learning Models:

Continue collecting data to refine the predictive models, enhancing their accuracy and applicability across different offender populations and crime types.

3

Integrate into Sentencing Guidelines:

Incorporate NCRA results into judicial decision-making processes to support more tailored and evidence-based sentencing and rehabilitation programs.

About the Study

The study was conducted by researchers at The Center for Science and Law in Houston, Texas, and Stanford University School of Medicine, under the leadership of David M. Eagleman. It represents an effort to leverage advances in neuroscience and machine learning to improve the criminal justice system’s approach to managing recidivism risk. The full APA citation is: Haarsma, G., Davenport, S., White, D. C., Ormachea, P. A., Sheena, E., & Eagleman, D. M. (2020). Assessing risk among correctional community probation populations: Predicting reoffense with mobile neurocognitive assessment software. Frontiers in Psychology, 10, 2926. https://doi.org/10.3389/fpsyg.2019.02926