I guess that you refer to processes controlled with automatic control systems (feedback). In this case I always use the integral of square error over time and the integral of square error of control signal oven time. Like I assumed that the processes have a feedback automatic control the target variable must be monitored and the set point for this target variable must be known. Therefore it is possible to evaluate the square of difference (monitored value-set point) during time and therefore the integral over time may be evaluated by numerical integration. Similarly, the control handle variable may be monitored during time and its square may be integrated over time. Both integrals must be minimal in the better control. However these integrals are competitive and the better control would be the control with the better equilibrium of two integrals (Nash equilibrium or Pareto front).
Dear Miguel, thank you for your response. I thought in terms of dual-process theories involving more reflexive or conscious (controlled) processing to planning and produce behavior and automatic (unconscious) processing. Some references that I thought:
Schneider, W., and Shiffrin, R. M. (1977). Controlled and automatic human information processing: I. Detection, search, and attention. Psychol. Rev. 84: 1–66.
Kahneman D. (2011). Thinking, Fast and Slow. London: Penguin Books.
I will read your article, who knows I have some interesting insights.
Excuse me Prof. Lage. I am engineer, and therefore when I read process controlling, I think in chemical engineering process. By the other hand, I am thinking that the integral over a given time of control action square for controlled human actions vs the integral over the same given time of control action square for autonomous human actions could be a feasible way to compare. The problem of course, if that would be necessary to identify variables defined in real numbers on a normalized metric that represent the human behavior actions. Maybe this metrics has been defined yet, really I do not know, but Norbert Wiener’s cybernetics is in that sense.
Thank you again for share your knowledge Prof. Miguel. We have usually used some motor task to infer about motor control and learning and we have defined some normalized metrics that better represents motor behavior. I think about your suggestion. Thank you!
what immediately came to my mind was the dual-tasking paradigm, maybe in conjunction with assessing concomittant neurological variables (.eg. fMRT-data), enegetic resource requirements of the central nervous system (eg. central nervous activation level, possibly assessed by means of flicker fusion frequency), and/or effort.
Then, of course, any specific choice of paradigm will depend on the (a) type of (motor) task in question (eg. sequence learning, reaction time task, precise force production; etc....), and therewith (b) the most task- and problem-appropriate theoretical concept (eg. capacity sharing; central bottleneck models; working memory dependent interference (i.e. Keele et al., 2003); type of movement representation (cf. Hikosaka), and so forth.
Thus I don't think there is a general "best answer". So, what kind of conceptual background and/or task requirements do you actually have in mind? Could you be a mite more specific, please?
May be, we can adjoin on a skype-session, and start talking this over?