Adaptive Scheduling for Time-Triggered Network-on-Chip-Based Multi-Core Architecture Using Genetic Algorithm

Adaptation in time-triggered systems can be motivated by energy efficiency, fault recovery, and changing environmental conditions. Adaptation in time-triggered systems is achieved by preserving temporal predictability through metascheduling techniques. Nevertheless, utilising existing metascheduling schemes for time-triggered network-on-chip architectures poses design time computation and run-time storage challenges for adaptation using the resulting schedules. In this work, an algorithm for path reconvergence in a multi-schedule graph, enabled by a reconvergence horizon, is presented to manage the state-space explosion problem resulting from an increase in the number of scenarios required for adaptation. A meta-scheduler invokes a genetic algorithm to solve a new scheduling problem for each adaptation scenario, resulting in a multi-schedule graph. Finally, repeated nodes of the multi-schedule graph are merged, and further exploration of paths is terminated. The proposed algorithm is evaluated using various application model sizes and different horizon configurations. Results show up to 56% reduction of schedules necessary for adaptation to 10 context events, with the reconvergence horizon set to 50 time units. Furthermore, 10 jobs with 10 slack events and a horizon of 40 ticks result in a 23% average sleep time for energy savings. Furthermore, the results demonstrate the reduction in the state-space size while showing the trade-off between the size of the reconvergence horizon and the number of nodes of the multi-schedule graph.

Time-based task expectancy: perceptual task indicator expectancy or expectancy of post-perceptual task components?

The temporal predictability of upcoming events plays a crucial role in the adjustment of anticipatory cognitive control in multitasking. Previous research has demonstrated that task switching performance improved if tasks were validly predictable by a pre-target interval. Hence, far, the underlying cognitive processes of time-based task expectancy in task switching have not been clearly defined. The present study investigated whether the effect of time-based expectancy is due to expectancy of post-perceptual task components or rather due to facilitation of perceptual visual processing of the coloured task indicator. Participants performed two numeric judgment tasks (parity vs. magnitude), which were each indicated by two different colours. Each task was either more or less frequently preceded by one of two intervals (500 ms or 1500 ms). Tasks were indicated either by colours that were each more frequently (or in Exp. 1 also less frequently) paired with the interval or by colours that were equally frequent for each interval. Participants only responded faster when colour and task were predictable by time (expected colour), not when the task alone was predictable (neutral colour). Hence, our results speak in favour of perceptual time-based task indicator expectancy being the underlying cognitive mechanism of time-based expectancy in the task switching paradigm.

Time for Action: Neural Basis of the Costs and Benefits of Temporal Predictability for Competing Response Choices

The brain can anticipate the time of imminent events to optimize sensorimotor processing. Yet, there can be behavioral costs of temporal predictability under situations of response conflict. Here, we sought to identify the neural basis of these costs and benefits by examining motor control processes in a combined electroencephalography–EMG study. We recorded electrophysiological markers of response activation and inhibition over motor cortex when the onset-time of visual targets could be predicted, or not, and when responses necessitated conflict resolution, or not. If stimuli were temporally predictable but evoked conflicting responses, we observed increased intertrial consistency in the delta range over the motor cortex involved in response implementation, perhaps reflecting increased response difficulty. More importantly, temporal predictability differentially modulated motor cortex activity as a function of response conflict before the response was even initiated. This effect occurred in the hemisphere ipsilateral to the response, which is involved in inhibiting unwanted actions. If target features all triggered the same response, temporal predictability increased cortical inhibition of the incorrect response hand. Conversely, if different target features triggered two conflicting responses, temporal predictability decreased inhibition of the incorrect, yet prepotent, response. This dissociation reconciles the well-established behavioral benefits of temporal predictability for nonconflicting responses as well as its costs for conflicting ones by providing an elegant mechanism that operates selectively over the motor cortex involved in suppressing inappropriate actions just before response initiation. Taken together, our results demonstrate that temporal information differentially guides motor activity depending on response choice complexity.

A between-task consequence of temporal expectations

In everyday life, we often anticipate the timing of one upcoming task or event while actively engaging in another. Here, we investigated temporal expectations within such a multi-task scenario. In a visual working-memory task, we manipulated whether the onset of a working-memory probe could be predicted in time, while also embedding a simple intervening task within the delay period. We first show that working-memory performance benefitted from temporal predictability, even though an intervening task had to be completed in the interim. Moreover, temporal expectations regarding the upcoming working-memory probe additionally affected performance on the intervening task, resulting in faster responses when the memory probe was anticipated early, and slower responses when the memory probe was expected late, as compared to when it was temporally unpredictable. Because the intervening task always occurred at the same time during the memory delay, differences in performance on this intervening task are attributed to a between-task consequence of temporal expectation. Thus, we show that within multi-task settings, knowing when working-memory contents will be required for an upcoming task not only facilitates performance on the associated working-memory task, but can also influence the performance of other, intervening tasks.

Confirmation of age-related alterations in inhibitory control using a modified minimally delayed oculomotor response (MDOR) task

Considerable effort has been made to measure and understand the effects of ageing on inhibitory control using a range of behavioural tasks. In the minimally delayed oculomotor response (MDOR) task, participants are presented with a simple visual target step with variable target display duration (TDD), and instructed to saccade to the target not when it appears (a prosaccade response), but when it disappears (i.e., on target offset). Using this task, we recently found higher error rates and longer latencies for correct responses in older compared to younger participants. Here we have used a modified MDOR task, in which participants were presented with static placeholders identifying potential target positions (increasing spatial information), and three TDDs rather than two (reducing temporal predictability). We found that the yield of analysable trials was generally higher with this modified task and in 28 older (mean ± SD age: 65 ± 7 y) and 25 younger (26 ± 7 y) participants the total overall error rate was again higher in the older group (30 ± 18% vs. 16 ± 11%). An analysis of the temporal distribution of responses demonstrated a pronounced peak in error production around 150 ms (young) or 200 ms (old) after target onset. When we recalculated the error rate focusing on these errors, it was again significantly higher in the older group. The latency of correct responses (to offsets) was significantly increased in the older group, although much of this increase was accounted for by expected age-related visuomotor slowing. However, both latency and distribution data suggested that while older participants could generate increased levels of inhibition, they could not maintain these levels as efficiently as the younger participants. In 24 participants (15 old, 9 young) who completed both versions of the MDOR task, neither latency nor error rates differed significantly between versions. These results confirm an inhibitory control deficit in healthy older participants, and suggest that the dynamics of inhibitory control are also affected by ageing. The modified MDOR task yields more data while not altering basic performance parameters.

Shielding working-memory representations from temporally predictable external interference

Protecting working-memory content from distracting external sensory inputs and intervening tasks is a ubiquitous demand in daily life. Here, we ask whether and how temporal expectations about external events can help mitigate effects of such interference during working-memory retention. We manipulated the temporal predictability of interfering items that occurred during the retention period of a visual working-memory task and report that temporal expectations reduce the detrimental influence of external interference on subsequent memory performance. Moreover, to determine if the protective effects of temporal expectations rely mainly on distractor suppression or also involve shielding of internal representations, we compared effects after irrelevant distractors that could be ignored vs. interrupters that required a response. Whereas distractor suppression may be sufficient to confer protection from predictable distractors, any benefits after interruption are likely to involve memory shielding. We found similar benefits of temporal expectations after both types of interference. We conclude that temporal expectations may play an important role in safeguarding behaviour based on working memory – acting, at least partly, through mechanisms that include the shielding of internal content from external interference.

Action-based predictions affect visual perception, neural processing, and pupil size, regardless of temporal predictability

Sensory consequences of one’s own action are often perceived as less intense, and lead to reduced neural responses, compared to externally generated stimuli. Presumably, such sensory attenuation is due to predictive mechanisms based on the motor command (efference copy). However, sensory attenuation has also been observed outside the context of voluntary action, namely when stimuli are temporally predictable. Here, we aimed at disentangling the effects of motor and temporal predictability-based mechanisms on the attenuation of sensory action consequences. During fMRI data acquisition, participants (N = 25) judged which of two visual stimuli was brighter. In predictable blocks, the stimuli appeared temporally aligned with their button press (active) or aligned with an automatically generated cue (passive). In unpredictable blocks, stimuli were presented with a variable delay after button press/cue, respectively. Eye tracking was performed to investigate pupil-size changes and to ensure proper fixation. Self-generated stimuli were perceived as darker and led to less neural activation in visual areas than their passive counterparts, indicating sensory attenuation for self-generated stimuli independent of temporal predictability. Pupil size was larger during self-generated stimuli, which correlated negatively with blood oxygenation level dependent (BOLD) response: the larger the pupil, the smaller the BOLD amplitude in visual areas. Our results suggest that sensory attenuation in visual cortex is driven by action-based predictive mechanisms rather than by temporal predictability. This effect may be related to changes in pupil diameter. Altogether, these results emphasize the role of the efference copy in the processing of sensory action consequences.