ReutersCan a tool or technology be applied to the brain and accurately predict out of a given group of people who will be the smartest?
The research arm of the Office of the Director of National Intelligence, Intelligence Advanced Research Projects Activity (IARPA) is looking for exactly those kinds of tools.
“IARPA is looking to get a handle on the state of the art in brain-based predictors of future cognitive performance. In particular, IARPA is interested in non-invasive analyses of brain structure and/or function that can be used to predict who will best learn complex skills and accomplish tasks within real-world environments, and with outcome measures, that are relevant to national security.
+More on Network World: 7 high-tech programs that want to pick your brain+
While conventional measures such as academic achievement, pencil-and-paper exams, and previous experience can be informative for predicting future performance, some recent research suggests that it may be possible to supplement (or supplant) traditional evaluation tools with direct measurements of the brain to confer additional predictive power. However, the extent to which neural tools can improve prediction of performance and expertise over and above more conventional tests remains in question, and the extent to which neural tools and measures have been tested using meaningful real-world outcome measures is also unclear, IARPA stated.
IARPA said with the program dubbed Neural Tools to Augment Prediction of Performance, Expertise, and Domain-knowledge (UNTAPPED), it is looking for insight into what it calls “credible next-generation tools, methods, and/or analyses that may overcome current technical and/or practical hurdles in predicting future cognitive performance, or that have been developed for other applications (such as predicting or assessing mental illness or psychological disorders) but have not yet been tested for this application.”
The researchers say that many organizations - from universities and companies to competitive sports teams and the military - are interested in accurately predicting an individual’s future cognitive performance and potential for different domain-knowledge and expertise.
“Such a predictive capability would allow organizations to determine in advance who is most likely to be able to learn and master complex skills and accomplish tasks in real-world environments that are important for the organization’s mission and success, thereby increasing return on investment for training activities and optimizing matching of personnel to tasks/environments. Such a capability would be particularly valuable in professions where job demands and required skills change rapidly due to new environments, new competitors, and/or advances in tools and methods. Indeed, given the accelerating rate of technical and social change impacting many organizations, as well as increasing costs for training and sustaining human resources, the ability to improve accuracy in predicting future cognitive performance for even a small percentage of current or future personnel could be significant.”
To get an idea of what IARPA is looking for, its UNTAPPED Request For Information includes the kinds of questions it is looking to answer with this new research, including:
1. Measures
a. What characteristics of brain structure and/or function are (or could be) used to predict future cognitive performance or potential expertise in different domains?
-
Which technologies are used to collect this information?
-
What are the experimental protocol(s) employed?
-
Have the measures been used in conjunction with a specific intervention (e.g.collected prior to or in parallel with task-related training to predict training outcome)?
-
e. What are the key technical or logistical challenges in collecting these data (e.g. multiple samples are needed per person and/or overall; extensive time is required to collect the data; imaging requires dye injection; analysis requires supercomputer; etc.)?
-
Which of these challenges (if any) are likely to be resolved with advances in primary or enabling technologies?
2. Outcomes
a. What types of cognitive performance have been (or could be) predicted with these measures? If this performance isn’t demonstrated in terms of real-world outcomes, is there evidence in the literature that these predictions could affect real-world outcomes?
b. For which types of cognitive tasks are neural measures likely to offer the most predictive power relative to more easily assessable behavioral predictors (e.g., pen and paper tests)?
c. How far in advance have these measures been shown to be predictive (i.e. do they predict performance one hour/day/week/year from when they are collected)?
d. What data support the finding that the measures can be used to predict performance in the future vice simply correlating with previous or current performance?
e. In what kind and size of populations have these results been demonstrated? How much variance is exhibited in the accuracy of prediction? How much additional variance could be expected by increasing the diversity of the subject population in terms of sex, age, IQ, occupation, nationality, culture, etc.?
f. What support, if any, have these measures received in peer-reviewed
literature? Have there been any reports (anecdotal or published) of negative findings using the same types of measures?3. Alternatives
a. Have these brain-based predictors been compared to or combined with conventional (non-neural) measures? If so, what are the relative costs/benefits? If not, against which conventional measures should the proposed approach be tested?
4. Limitations
a. Given that performance on a complex cognitive task or expertise in a specific domain may be mediated by a host of intermediate factors, what are likely to be fundamental limitations (theoretical and/or practical) to the development, testing, or use of neural tools and measure(s)?
This brain research is just one of many ongoing projects IARPA has going. In July the group talked about a project it says could revolutionize machine intelligence by constructing algorithms that utilize the same data representations, transformations, and learning rules as those employed and implemented by the brain.
The specific goal of that program, known as Machine Intelligence from Cortical Networks (MICrONS) is to create what IARPA calls “a new generation of machine learning algorithms derived from high-fidelity representations of the brain’s cortical microcircuits to achieve human-like performance on complex information processing tasks.
+More on Network World: Security Holy Grail anyone?; US sets $1.4M to get unique metaphor-recognizing software system humming+
In March, IARPA announced the Investigating Novel Statistical Techniques to Identify Neurophysiological Correlates of Trustworthiness (INSTINCT) Challenge where it asked members of the public to develop algorithms that improve predictions of trustworthiness, using neural, physiological, and behavioral data recorded during experiments in which volunteers made high-stakes promises and chose whether or not to keep them.
Follow Michael Cooney on Twitter: nwwlayer8 and on Facebook.
Check out these other hot stories:
MIT system will make oxygen on next NASA Mars mission
Federal court system warns of new email jury scamRocket Lab wants to make Model T of space satellite launchers
FTC urges mobile carriers to help send cramming charges packing
NASA looking for out-of-this-world Mars communications services
FTC takes out “tech support” scammers; $5.1 million in fines, retribution
Finding life in space by looking for extraterrestrial pollution