6. Conclusions
6.2. Limitations and Future Research
There were some limitations to this study related to the survey tool and performance metrics. More work could be done to further refine the questions in the biometric user profiling
response could be used for questions inquiring about regularity of behavior, such as consuming caffeinated items, using a computer (in the past or present), engaging in meditation, and drinking alcohol. For example, without a scale the question about regular alcohol consumption may have been interpreted more widely than intended; some participants may have considered themselves to regularly drink alcohol if they consumed a drink each weekend whereas some may have consumed a drink multiple times in a day. Although this question was not attempting to reveal alcoholism, there may be related physiological implications to consider for people classified as alcoholics versus “social drinkers”.
For performance metrics, participants attempted a difficult task to attain a bounded target instead of simply activating the transducer above or below a certain threshold to attain an infinitely large target area. Performance should increase significantly with a thresholded task. Further, although detrending was not used in this study, participants should benefit from use of an interface that dynamically calibrates because sometimes peoples’ ranges shifted over the course of one task. However, this study was not focused on understanding the interface designed, but rather on understanding the underlying relationships of individual characteristics with control.
Further examination of individual characteristics and the potential differences between able- bodied and disabled participants’ motivation and frustration levels is still necessary. Also, a larger sample size would allow stronger conjectures by using parametric statistics for data analysis. Unfortunately, achieving a larger sampling of disabled participants poses a definite challenge. According to members of the ALS Association, although approximately 30,000 ALS patients reside in the United States many of them are unaware of support networks available and do not have a technically-savvy team to help connect them with resources. Therefore, it may be
necessary to recruit participants with illnesses and conditions other than ALS that result in motor disabilities. In addition, further investigation into the effects of handedness on performance may prove interesting because of differences in cognitive development.
This study focused on initial control but does not specifically address the potential long term effects of training with biometric interface technologies on performance. It may be possible that training could supersede characteristics from a person’s biometric user profile. A longitudinal, controlled study is needed to isolate the effects of training on performance with particular biometric interface technologies. In particular, a fully-crossed (2 x 2) free experiment could be run to compare people who are predicted to fit either a fNIR or GSR biometric interface
technology with those who are predicted not to fit and people in both of these groups who have received training with those who have not. Able-bodied participants may be used to obtain a larger sample size.
Although tested with a select set of biometric interface technologies, we may expand this methodology to explain other types of continuous input technology, such as some EEG-based recordings, and the effect of task on performance. In the future, this work may be extended to technologies with discrete types of transducer output, tested under different conditions, resulting in a more widely validated model that may be used for prediction of performance. Although not addressed here, future work could explore the temporary effects of drugs and caffeine on
performance or differences in environments that impose temporary disabilities. In addition to testing with different biometric interface technologies, different tasks may be analyzed for their effect on performance. Only one task was investigated here, but a study to replicate and extend this work may reveal a deeper cognitive relationship between the task and the type of biometric interface technology.
This work may be extended into a protocol for conducting integrated remote assessments for users of biometric interface technologies. Results of this dissertation work may serve as
screening and analysis portions of a protocol which can be shared across remote teams to determine what system works best for a user. However, because an integrated approach for remote team collaboration would likely require a form of high-computing infrastructure such as grid-technology and deeper investigation into related telemedicine techniques, this is presently outside the scope of a dissertation but may be appropriate for career work.
Finally, opportunities exist for future collaborations with governmental organizations, such as the Air Force, to understand if biometric interface technologies like fNIR and GSR may be used in conditions with extreme forces that affect vascular activity. Performance may be
compared for fNIR, GSR, and EEG devices, where EEG has previously been investigated for use by jet pilots. This represents just one of many directions for future research as a result of this work.