OBJECTIVE: The objective of this topic is to develop an automated mental workload classifier sensitive to systems and tasks, as well as diagnostic of different types of mental workload experienced. DESCRIPTION: The measurement of Soldier mental workload during test and evaluation (T&E) is critical to understanding total system (comprised of the human and system) performance issues. Currently the measurement of mental workload is a two step process. First, a training session is conducted prior to the actual test event. The training session is used to build a model of each subject. During the training session, subjects are instrumented to acquire physiological data while they perform specific cognitive tasks having pre-assigned high and low states. The physiological data is then processed and psycho-physiological features of the subject"s mental state are extracted and used to build a linear model. Once the models are developed and trained, the actual test event can begin. During the test event, the model classifies new physiological data acquired from the subject while they perform specific tasks associated with a system. This approach provides a linear scale of the mental workload experienced by subjects. However, the approach is limited by only being able to determine when the mental workload was experienced and at what the level (high or low). Assessing total system performance issues with the current capability is challenging, if not impossible. A more comprehensive measure of mental workload is necessary for T&E. The T&E community desires a mental workload measurement that is sensitive to the task or system and capable of diagnosing the type of mental workload experienced. A sensitive mental workload measure is needed to discriminate between the workload imposed by one system or task versus the workload imposed by another system or task. A diagnostic mental workload measure is required to discriminate between different types of mental workload (visual, auditory, cognitive, fine motor, gross motor, speech, and tactile) experienced during a task or activity. Such a capability will enable testers and developers to optimize mental workload by identifying when a particular type of mental workload occurred and the system component or task associated with that mental workload. The ideal mental workload classifier would resemble the US Army Research Laboratory (ARL) Improved Performance Research Integration Tool (IMPRINT) mental workload model. The IMPRINT model decomposes mental workload into 7 types: visual, auditory, cognitive, fine motor, gross motor, speech, and tactile. These 7 types of mental workload are further decomposed into behaviors. Each behavior is associated with a numeric value. As the behaviors for each type of workload become more complex, their numeric value increases. This establishes an ordinal and interval scale of mental workload. The IMPRINT model also takes into account the fact that several mental workload types may be used simultaneously, thus creating conflict which impacts the overall mental workload experienced. Finally, the IMPRINT model considers strategies to mitigate or compensate for high workload. By combining these features, the IMPRINT model can predict subject mental workload over time, predict the type of mental workload experienced, and identify the task or system component associated with the mental workload. This powerful tool enables the optimization of mental workload for system design and analysis. The intent of this SBIR is to develop a physical capability to replicate what IMPRINT can do in the modeling world. The desired outcome from this topic is a mental workload classifier for physiological data built around the IMPRINT model of mental workload. The classifier shall be capable of discriminating between mental workload types and associated behaviors. Additionally, the classifier shall be capable of accounting for conflicts between different types of workload. The desired product is a software application, and the associated source code, capable of operating on a Windows XP platform or greater. PHASE I: The Phase I project shall determine the technical feasibility for developing a mental workload classifier based on the IMPRINT workload model. The feasibility study shall focus on the development of a classifier capable of discriminating between various types of workload. Phase I shall also include the design of an IMPRINT based mental workload classifier and the development of approaches for implementing the classifier. Phase I deliverables will include results from the feasibility study, the initial classifier design concept, initial development approach, monthly progress reports and a final report. The Phase I proposal shall emphasize the breadth of the commercial market for the proposed classifier and explain in detail how the proposed classifier will fit into the commercial market. PHASE II: The Phase II project shall develop a prototype automated mental workload classifier and the associated methodology for implementing the classifier. Efforts during Phase II shall also include a validation study to ensure classifier results. The study shall compare classifier results against IMPRINT predicted workload for a specified task. Phase II deliverables include the software application for the mental workload classifier, source code for the classifier application, a technical manual on how to use the classifier, results of the classifier validation study, monthly progress reports, and a final report. PHASE III: The vision for this research effort is to develop a fully automated mental workload classifier and methodology for use in real time monitoring of subjects. This effort has the potential to transition to various military applications. One specific application is the Natick Soldier Center (NSC) Soldier Load Management program. The foundation for this program is to optimize the Soldier"s physical and mental load. The classifier developed from this SBIR topic can play a vital role in the design of systems for the Soldier Load Management program and it can be embedded as a part of the fundamental technologies within the program to assess Soldier mental workload. Additional transition paths include the Test Resource Management Center (TRMC) Tri-Service Warfighter Performance Test Capability (TSWPTC) program. TSWPTC is aimed at identifying and developing critical Warfighter performance testing capabilities across DOD agencies. The program kicked off with a study which identified and prioritized gaps associated with Warfighter performance test capabilities. The number two gap identified was physiological state, which includes mental workload. The Tactical Human Integration with Networked Knowledge (THINK) and High Definition Cognition ATOs are possible transition paths for this SBIR effort. These ATOs will benefit from the advanced mental state classification algorithms developed under this SBIR effort. The classifications algorithms have the potential to be applied to other psycho-physiological measurements for more in-depth understanding of neuro-cognitive process and aid in development of neuro-ergonomically designed Soldier-system interfaces. Commercial transition paths include the growing brain computer interface (BCI) and ergonomics communities. With respect to the BCI community it is envisioned that the classifier developed from this effort will be easily adaptable to other brain measures or functions. This will allow BCI developers the opportunity to quickly tailor the classifier to meet their system needs. Additionally, the classifier has great potential for transitioning into the gaming market. There are a variety of systems currently being sold whereby users interact with a video game or even a toy via their brain waves. With the advanced classifier developed from this SBIR, developers can greatly enhance the types of mental interactions possible. With respect to the ergonomics community there is a large interest in the assessment of operator performance based on ergonomic design concepts. Mental workload measures related to operator performance are of great interest to the ergonomics community, particularly in the automotive and aviation fields. Markets for the development and enhancement of air traffic controller technologies are a prime use case for highly sensitive and diagnostic mental workload classifiers.