Systems factorial technology

Current technology and workplace environments are designed to enable people to attempt multiple tasks simultaneously. Consequentially, people divide their limited attentional resources among many competing demands. In some recent work, Morey et al. (2018) found the limited processing capacity to redundant targets in a peripheral task did not change depending on the difficulty or presence of a dual-task. Nonetheless, it is unknown how 1) the introduction of, or increased difficulty of, a second task may change how people combine multiple peripheral targets (e.g., in parallel or serial) and 2) processing efficiency may depend on both the salience of peripheral targets and the presence/difficulty of a second task. In this work, we use systems factorial technology to investigate the cognitive processing mechanisms of redundant visual targets in isolation and in the context of an easy or difficult multiple object tracking (MOT) task. We manipulate the degree of MOT demands: track 0, 1, or 4 dots, and the salience of peripheral red target squares (easy, difficult). We find limited-capacity parallel-OR processing of redundant targets but the degree of limited processing capacity depends on the demands of the MOT (0, 1, or 4 dots) and the salience of the dual-targets (low/high). Our data suggest that the structure for how people process multiple peripheral cues does not change depending on the overall attentional demands of the task(s); however, the extent that people benefit from redundant information may depend on how difficult it is to perceive the targets and external task demands.

Understanding how human performance changes as the amount of information available varies is of particular interest across many basic and applied research topics in psychology. One approach to quantifying these changes is with the assessment functions. Briefly, the assessment functions are a family of non-parametric measures that compare observed performance to a baseline derived from a model predicting how changes in information influence the system. Although less commonly used than the capacity coefficient, a similar measure based only on response-time data, the assessment functions are a promising tool because it accounts for response-time and accuracy, and hence is applicable in conditions in which speed-accuracy trade-offs can vary. Two potential hinderances to the wider use of the assessment functions are the specific assumptions needed to derive the baseline model and the lack of associated inferential statistics. In this talk, we demonstrate how a fixed accumulator model with a random threshold (i.e., Grice model) representation of the choice/RT data can be leveraged to derive generalized assessment functions and, potentially, for deriving inferential statistical tests.

Research on the conjunction fallacy has largely been concerned with exploring its underlying causes. Conventional methods primarily use descriptive scenario-based tasks to represent probabilities. However, such descriptive methods are prone to misinterpretation and other cognitive biases. We attempt to demonstrate how these problems can be overcome by applying a psycho-physical framework, which represents probabilities as proportions of blue-to-orange squares in a grid. Results show that a psycho-physical framework can demonstrate a conjunction fallacy. Preliminary evidence accumulation modelling shows that the conjunction fallacy appears to be driven by changes in information processing specifically and not by changes in speed-accuracy trade-offs. To further investigate the information processing involved in the conjunction fallacy, systems factorial technology was applied to the results. This was done by fitting a linear ballistic accumulator to the empirical results, then simulating experimental responses and response times based on the logic of systems factorial technology information processing architectures. Simulation results go on to show that because the inherent logic of the systems factorial technology architectures assume independent processing of information in the different channels, they cannot produce conjunction fallacies. However, allowing crosstalk between the architecture channels, either via a bias in the start point of evidence accumulation for serial architectures, or collapsing the response threshold after terminating evidence accumulation in one channel for parallel architectures, can lead to conjunction fallacies. Empirical and simulation results are argued to show that a bias in the start point of evidence accumulation for serial architectures is the most plausible cause of conjunction fallacies.

Revolving around a two-stage decisional paradigm where a categorical decision was followed by an action decision, an inconsistency in choice behavior when both decisions were explicitly measured versus when only the second decision was measured has been revealed and replicated in the past twenty years. Such an inconsistency in choice behavior, referred as the interference effect, violates the fundamental properties of probability theory: the law of total probability and the Markov property and thus challenges a wide range of classical cognitive models of decision-making. Substantial theoretical efforts in the past decade have been devoted to interpreting the underlying cognitive mechanisms producing the interference effect. However, most of these efforts have relied on critical assumptions of the underlying cognitive structure and did not consider the response-time performance. To this end, the current study adapted the two-stage decisional paradigm for the extended application of a set of theory-driven response-time based measurements. Conjoining the utilization of the response-frequency measurement, we probe the underlying cognitive properties that may relate to the occurrence of interference effect. The results showed that with fewer restrictions on a sequential processing order of categorization and action decisions, the underlying cognitive systems tended to follow a parallel mental architecture and the processing speed of the deliberation processes tended to facilitate each other, along with observations of the interference effect in response frequencies. These results suggested that interference effects might be closely pertinent to cognitive systems characterized by parallel mental architecture and positively interact underlying deliberation processes of categorization and action decision.

Previous studies have shown individuals with higher level of aerobic fitness exhibited better cognitive control. However, less is known about how aerobic fitness level relates to resilience capacity, a measure of the change in multi-signal processing efficiency in the presence of the distractors. Thus, the aim of the present study is to examine whether aerobic fitness is related to individual differences in resilience capacity.Twenty-two young adults with higher level of aerobic fitness (high-fit group; aged 21.05 ± 2.15 years; VO2max = 58.36 ± 6.71 ml/kg/min) and twenty-two demographically matched lower aerobic fitness counterparts (low-fit group; aged 22.23 ± 1.38 years; VO2max = 41.74 ± 4.03 ml/kg/min) performed a Go/Nogo version of the redundant-target detection task. According to Systems Factorial Technology (SFT), resilience capacity was assessed by comparing the processing efficiency when two targets were simultaneously presented to when a target and a distractor were presented. Further, a functional principal component analysis (fPCA) was applied for exploratory analysis of the resilience capacity. Results revealed no group differences in mean reaction times (RTs) across task conditions. In terms of SFT, the fPCA results revealed larger resilience capacity in the high-fit group compared with the low-fit group for the faster responses, while such difference was not found for the slower responses. Novel to the current study is to provide a more comprehensive investigation of the cognitive benefits of aerobic fitness. In conclusion, the study suggests that the beneficial association of greater aerobic fitness with information processing efficiency may change dynamically across response times.