An adaptive network model covering metacognition to control adaptation for multiple mental models

Learning processes can be described by adaptive mental (or neural) network models. If metacognition is used to regulate learning, the adaptation of the mental network becomes itself adaptive as well: second-order adaptation. In this paper, a second-order adaptive mental network model is introduced for metacognitive regulation of learning processes. The focus is on the role of multiple internal mental models, in particular, the case of visualisation to support learning of numerical or symbolic skills. The second-order adaptive network model is illustrated by a case scenario for the role of visualisation to support learning multiplication at the primary school.     

Effects of training and display content on Level 2 driving automation interface usability

IntroductionLevel 2 driving automation features, such as adaptive cruise control (ACC) combined with lane centering, primarily communicate their operating statuses through the instrument cluster. It remains unclear how interface-specific training and display content influence the ability to understand Level 2 activity in production vehicles.MethodsEighty participants viewed videos recorded from the driver’s point of view in a variety of driving scenarios with Level 2 activity displayed in the instrument cluster of a 2017 Mercedes-Benz E-Class. Half viewed one of two instrument cluster layouts (simple or complex), and half received an orientation to the interface before the experiment. After each video, they were asked about the scenario they had just seen. We then examined what information in the displays participants used to identify Level 2 activity and the usability of the displays.ResultsTraining improved recognition accuracy of when lane centering was temporarily inactive and understanding of why it was inactive. Neither training nor display content affected the ability to recognize when ACC had adjusted the vehicle’s speed or detected a vehicle ahead, or when ACC initially did not detect a lead vehicle and understanding of why it had not detected it. Both factors influenced which sources of information participants used to determine Level 2 activity. Recognition accuracy of system activity improved when participants used valid sources of information. Training, but not display content, improved lane centering usability, but not ACC usability.DiscussionBasic training improves detection of notifications that potentially require further driver action, but not of those that display persistent status information. Training does not result in full understanding of all system notifications or functional limitations, which reinforces the need for intuitive, salient communication about system behavior and its limitations.     

Portrayal of personality in Victorian novels reflects modern research findings but amplifies the significance of agreeableness

All literature embodies an implicit theory of personality and human nature (Hogan, 1976). The research described here investigates the implicit personality theory embedded in the behavior of 435 characters in 143 canonical Victorian novels. Characters were rated on the Web by 519 scholars and students of 19th-century British literature. Ratings included the characters’ goals, success in achieving goals, mate preferences and strategies, and personality according to the Five Factor Model. Results suggest that novels by Victorian authors largely reflect personality and human nature as understood by modern personality psychology, but Victorian authors amplify the significance of agreeableness and thus, whether intentionally or not, encourage cooperative impulses in readers.     

Investigation of drivers’ thresholds of a subjectively accepted driving performance with a focus on automated driving

Lately, the development and implementation of automated driving moved to the center of interest in the automotive industry. In this context, one of the central issues – the configuration of adequate trajectories – is mainly tackled using a technical approach. However, it appears that a technically ideal driving performance does not necessarily coincide with the drivers’ subjective preferences. This study strives to determine thresholds of a subjectively accepted driving performance regarding lateral vehicle control. A second objective is to analyze the influence of selected personal and situational factors on these thresholds. An empirical online survey with 161 participants rating video sequences of driving performances was conducted. The video sequences differed not only with regard to the lateral offset of the ego-vehicle but also concerning the weather (sun/rain) and traffic conditions (existence/driving behavior of oncoming traffic). Additionally, the participants’ driving experience and sensation seeking were considered in the data evaluation. To analyze the data, binary logistic regression analyses were calculated. They revealed that the subjective evaluation of driving performances varies primarily depending on the lateral offset of both the ego-vehicle and the oncoming traffic. The results indicate that regarding the lateral offset certain thresholds of subjectively accepted driving performances do exist. Regarding the development of automated driving systems, two issues need to be considered in order to ultimately guarantee user acceptance. First, the subjective thresholds need to be integrated into the systems’ trajectory planning. Second, the oncoming traffic’s driving behavior has to be considered.     

Traction-energy balancing adaptive control with slip optimization for wheeled robots on rough terrain

On rough terrain, excessive wheel slippage is easily generated by changes of surface conditions such as soil types and geometries. It induces considerable loss of wheel traction and battery energy. To prevent this, wheeled robots should consistently recognize the current situation generated between wheel and surface. And also wheeled robots are required to optimally control wheel motion in limited wheel traction and battery capacity. Therefore, this paper proposes a novel wheel control algorithm based on slip optimization of traction and energy, which is adaptive to change of surface condition. Proposed wheel control algorithm is called Traction-Energy Balancing Adaptive Control (TEB) in this paper and TEB assigns optimized rotation speed to each wheel by observing wheel slip ratio which is a key parameter of TEB. As functions of TEB, TEB is largely divided into three main parts; (1) slip optimizer (2) slip controller (3) SC-compensator. In the slip optimizer, two optimal slip models were derived as a function of slip ratios regarding maximum traction and tractive efficiency using experimental data about wheel-terrain interaction in three types of soil (grass, gravel and sand). And the optimal slip models were employed in order to determine a desired slip value of wheel with observation of a change in actual robot velocity as control input in the slip controller. For optimal slip control, the proposed slip controller is based on conventional PID controller with compensating disturbance in the controller (SC-compensator) which occurs by change of surface shapes. In the SC-compensator, radial function networks (RBFN) was applied in the slip controller and RBFN was of help to readjust previously set PID gains depending on occurred slip error. Finally, TEB was experimentally verified by controlling a real robot having four wheels on various terrain types.     

Adaptive working memory training reveals a negligible effect of emotional stimuli over cognitive processing

Here we analyze how performance differences in an adaptive cognitive training regime based on the n-back task interact with emotional stimuli (scenes and faces) varying in their valence (negative, positive, and neutral). One hundred and three participants completed four training sessions across 2 weeks showing remarkable improvements from time to time. Results revealed similar results for faces and scenes regarding accuracy levels across increased complexity levels. However, reaction times (RTs) were sensitive to emotional conditions to some extent. Observed faster RTs to negative faces (disgust) were consistent with the negativity bias phenomenon, but this effect vanished for the highest levels of processing complexity. It is suggested that emotional information contents fail to interact with cognition when there are no cognitive resources left after the primary task is addressed.     

Estimation and visualization of confusability matrices from adaptive measurement data

We present a simple but effective method based on Luce’s choice axiom [Luce, R.D. (1959). Individual choice behavior: A theoretical analysis. New York: John Wiley & Sons] for consistent estimation of the pairwise confusabilities of items in a multiple-choice recognition task with arbitrarily chosen choice-sets. The method combines the exact (non-asymptotic) Bayesian way of assessing uncertainty with the unbiasedness emphasized in the classical frequentist approach.We apply the method to data collected using an adaptive computer game designed for prevention of reading disability. A player’s estimated confusability of phonemes (or more accurately, phoneme-grapheme connections) and larger units of language is visualized in an easily understood way with color cues and explicit indication of the accuracy of the estimates. Visualization of learning-related changes in the player’s performance is considered.The empirical validity of the choice axiom is evaluated using the game data itself. The axiom appears to hold reasonably well although a small systematic violation is observable for the smallest choice-set sizes.     

A Bayesian-optimal principle for learner-friendly adaptation in learning games

Adaptive learning games should provide opportunities for the student to learn as well as motivate playing until goals have been reached. In this paper, we give a mathematically rigorous treatment of the problem in the framework of Bayesian decision theory. To quantify the opportunities for learning, we assume that the learning tasks that yield the most information about the current skills of the student, while being desirable for measurement in their own right, would also be among those that are efficient for learning. Indeed, optimization of the expected information gain appears to naturally avoid tasks that are exceedingly demanding or exceedingly easy as their results are predictable and thus uninformative. Still, tasks that are efficient for learning may be experienced as too challenging, and the resulting failures can lower motivation. Therefore, in addition to quantifying the expected informational benefit for learning of any prospective task to be presented next, we also model the expected motivational cost of its presentation, measured simply as the estimated probability of failure in our example model. We propose a “learner-friendly” adaptation algorithm that chooses the learning tasks by optimizing the expected benefit divided by the expected cost. We apply this algorithm to a Rasch-like student model implemented within a real-world application and present initial results of a pilot experiment.     

Adapting to test structure: Letting testing teach what to learn

We propose that we encode and store information as a function of the particular ways we have used similar information in the past. More specifically, we contend that the experience of retrieval can serve as a powerful cue to the most effective ways to encode similar information in comparable future learning episodes. To explore these ideas, we did two studies in which all participants went through study–test cycles of single category lists while we manipulated the nature of the recognition tests. The recognition tests either included only same-category lures or only different-category lures. The experience of repeated testing leads participants to avoid conceptual-based strategies but only when conceptual knowledge was poorly diagnostic for recognition (i.e., in the same-category lures condition). In a second study with a similar manipulation, we showed that repeated testing with lures from the same category as study items improved performance in a final recall surprise test compared to conditions in which different-category lures were used. Such a difference is akin to the one obtained when encoding instructions focus on distinctive item features compared to cases in which the focus is on relational processing. We suggest that testing requirements lead to adaptive changes at encoding.