Goal of research
The goal of the current project is to study the brain mechanisms of decision making and performance monitoring in conditions of internal uncertainty.
Although internal uncertainty has long been known in scientific literature [Kahneman, Tversky, 1982], psychophysiological mechanisms of internal uncertainty during decision making remain largely understudied both in Russian and international literature.
The use of the concept of internal uncertainty allows broadening the theoretical paradigm of cognitive control and analyzing it with the help of psychophysiological indices of brain activity in tasks involving high level of cognitive load.
In our previous research [Novikov et al., 2017], we demonstrated that during performance on the task requiring high level of cognitive load, deficient information processing (during the analysis of the sensory stimulus or during establishing the correspondence between stimuli and responses) leads to the state of internal uncertainty. Importantly, we hypothesize that the state of internal uncertainty significantly differs from the state of conflict, the latter being a classical pivotal notion within the cognitive control paradigm.
Methods and empirical base of research
Execution of the current research project involved using original approaches towards induction and detection of the state of internal uncertainty. The study was based on four experimental factors related to internal uncertainty. Within the current project, in addition to the response time factor [Novikov et al., 2017], we introduced three new factors, which allowed us to study the state of internal uncertainty in three time intervals – at the stage of evidence accumulation and decision making, at the stage of initialization and implementation of the decision, and at the stage of expectation and processing of the feedback signal.
Having this as a purpose, we created a new modification of the condensation task, which was successfully used in our previous studies [Chernyshev, Medvedev, 2016; Novikov et al., 2015; Novikov et al., 2017], yet we introduced into this task several additional features of the experimental procedure that allowed us to analyze the factors of the internal uncertainty mentioned above. The condensation task is associated with high cognitive load, this feature distinguishing it from easier tasks commonly used in the studies of cognitive control. Thus, the condensation task is an adequate tool for studying internal uncertainty.
We recorded electroencephalogram and applied advanced mathematical methods of analysis.
Results of research
Corroborating our hypotheses, analysis of the psychophysiological indices revealed that the basic experimental factors of the current study (chimeric stimuli, long response time, uninformative feedback signal as well as motor hesitations before the actual behavioral response) indeed revealed association with the state of internal uncertainty.
For the first time, we have demonstrated that under the state of internal uncertainty P2 amplitude was increased. Moreover, this effect was clearly different from the effect of errors – since P2 amplitude did not depend upon trial outcome (correct or erroneous). In the state of internal uncertainty caused by chimerical stimuli, the effect of lowered P2 amplitude was accompanied by the effect of decrease (i.e. cancellation of increase) of theta oscillation in the pre-response time interval. This result cannot be explained within the traditional interpretation of post-stimulus theta oscillations being the correlate of conflict. On the contrary, this effect corroborates our point of view according to which the state on internal uncertainty has principal differences from the state of conflict: while conflict leads to the increase in post-stimulus and pre-response theta oscillations, the state of internal uncertainty does the opposite – it decreases theta oscillations at this state of decision making. Thus, the state of internal uncertainty, unlike the state of conflict, may be characterized by the lack of competition between the two alternative motor programs – because none of programs is sufficiently activated.
For the first time, we have demonstrated that in the condition of the uninformative feedback (thus preventing external error detection), activity of the system of the internal error detection increases, as evidenced by its correlates such as ERN/CRN. This evidences the possibility of appropriate adaptation within the cognitive control networks, leading to increased effectiveness of internal error detection in conditions when external error detection is not achievable.
For the first time, we have demonstrated that the state of internal uncertainty leads to the increase in the processes related to expectation of the feedback signal – as evidenced by the increase in at least two electrophysiological indices in the time window preceding the feedback onset (appearance of CNV and decrease of suppression of alpha oscillations).
Moreover, for the first time, we have demonstrated that, as predicted by our hypotheses, the power of alpha oscillations before the feedback reflects expectation of the feedback (i.e. it is anticipatory in nature) rather than it reflects a post-error adaptation that follows internal error detection.
For the first time, using the original method of detection of hesitant responses, we have demonstrated that the responses accompanied by such motor hesitations are actually responses that where initiated as errors but corrected "on the fly" (or at least attempted to be corrected). This is evidenced by increased Pe amplitude. On the whole, our findings evidence that Pe has direct relation to aware errors, as increased Pe was observed on trials involving motor hesitations before the choice of the response (when finger trajectory was recorded before pressing the button), and on trials involving slowed computer mouse movements (in conditions of mouse-tracking when subjects were to make responses by moving the mouse).
For the first time, we have demonstrated that in conditions of internal uncertainty, brain responses to the feedback signal were specifically dependent upon the type of the feedback (positive, negative or uninformative) – this was evidenced by changes in FRN amplitude, as well as changes in the power of alpha and beta oscillations.
Activation of theta oscillations after the positive feedback in the conditions of increased internal uncertainty can be explained by the relation of theta oscillations to unexpectedness, i.e. prediction error of the trial outcome ("unsigned" prediction error); that largely corroborates our hypotheses.
Novelty, importance, and practical applicability of the results
The novelty of the current project is related to the fact that it is the first complex psychophysiological study of the state of uncertainty in the context of the cognitive control paradigm. New results were obtained that have no direct analogs within Russian and international scientific literature.
The use of the notion of internal uncertainty allows broadening the theoretical concept of cognitive control and analyzing within its framework psychophysiological indices of brain activity in task with high cognitive load, which were created to better simulate challenges in professional activity. This opens prospects of new studies based on more realistic modeling of decision making.
The concept of internal uncertainty offers an important extension to the concept of decision making in comparison to the classical diffusion-drift model [Ratcliff, McKoon, 2008; Shadlen, Newsome, 2001], and to the classical understanding of the state of conflict [Botvinick et al., 2001; Yeung, 2014].
Importance of the current project is mostly fundamental in nature, as our study was aiming at understanding deep neurocognitive mechanisms that determine precision and speed of human performance, as well as flexibility and adaptivity of human behaviour. The study of brain mechanisms of cognitive processes is one of the high-priority interdisciplinary areas for the development of modern science.
Findings of the current studies may have importance for practically oriented research in the area of optimizing professional performance.