Cognitive Processing of Visually Presented Data in Decision Making.

OBJECTIVE: Intelligence analysis is a necessary operation for organizations engaged in adversarial decision making in military, criminal justice, and civil spheres. In particular, we focus on studying the sentiment of actors in a military-relevant decision-making task and how information pertaining to these actors is presented. BACKGROUND: Two intelligence analysis systems were created to represent sentiment data. METHOD: These systems and another with the raw information (i.e., the control condition) were presented to experimental participants in separate trials with the pretext that they act as intelligence analysts and compare two social media actors in U.S.-occupied territory on given subject matters. RESULTS: Results demonstrated that cognitive processing of the information improved and was preferred for the prototype that presented information in scatterplot form rather than a line graph, though side-by-side comparisons also were preferred from the disfavored system. CONCLUSION: We conclude with suggestions for designing intelligence analysis software that conforms to best practice in human factors with a focus on cognitive processing. APPLICATION: We hope these recommendations will improve intelligence analysis of sentiment information via a new system.

Editor's Review and Introduction: Cognition-Inspired Artificial Intelligence.

Cognitive science has much to contribute to the general scientific body of knowledge, but it is also a field rife with possibilities for providing background research that can be leveraged by artificial intelligence (AI) developers. In this introduction, we briefly explore the history of AI. We particularly focus on the relationship between AI and cognitive science and introduce this special issue that promotes the method of inspiring AI development with the results of cognitive science research.

Underestimating numerosity of items in visual search tasks.

Previous research on numerosity judgments addressed attended items, while the present research addresses underestimation for unattended items in visual search tasks. One potential cause of underestimation for unattended items is that estimates of quantity may depend on viewing a large portion of the display within foveal vision. Another theory follows from the occupancy model: estimating quantity of items in greater proximity to one another increases the likelihood of an underestimation error. Three experimental manipulations addressed aspects of underestimation for unattended items: the size of the distracters, the distance of the target from fixation, and whether items were clustered together. Results suggested that the underestimation effect for unattended items was best explained within a Gestalt grouping framework.

The Effect of Traumatic Brain Injury (TBI) on Cognitive Performance in a Sample of Active Duty U.S. Military Service Members.

INTRODUCTION: Traumatic brain injury (TBI) is considered a signature injury from the fighting in Iraq and Afghanistan. Since the year 2000, over 370,000 U.S. active duty service members have been diagnosed with TBI. Although prior research has shown that even mild forms of TBI are associated with impaired cognitive performance, it is not clear which facets of cognition (computation, memory, reasoning, etc.) are impacted by injury. METHOD: In the present study, we compared active duty military volunteers (n = 88) with and without TBI on six measures of cognition using the Automated Neuropsychological Assessment Metric software. RESULTS: Healthy volunteers exhibited significantly faster response times on the matching-to-sample, mathematical processing, and second round of simple reaction time tasks and had higher throughput scores on the mathematical processing and the second round of the simple reaction time tasks (P < 0.05). CONCLUSION: In this population, cognitive impairments associated with TBI influenced performance requiring working memory and basic neural processing (speed/efficiency).

Effect of pacing and working memory loads on error type patterns in a routine skill.

Event counting provides a laboratory paradigm for studying types and causes of error in routine activity. Experiment 1 demonstrated that as more time per event is allowed, counting errors typically are undercounts, then overcounts as the time per event is extended to approximately 3 s. Experiments 2 and 3 examined 2 possible causes of this phenomenon, forgetting and confusions due to overrehearsal of the next number to be used. These findings demonstrate that overcount errors during event counting are reduced by working memory loads, which may result from a lower ability to rehearse the current total. Implications for theoretical accounts of goal representation, control, and error monitoring are discussed.

Simulations in Cyber-Security: A Review of Cognitive Modeling of Network Attackers, Defenders, and Users.

Computational models of cognitive processes may be employed in cyber-security tools, experiments, and simulations to address human agency and effective decision-making in keeping computational networks secure. Cognitive modeling can addresses multi-disciplinary cyber-security challenges requiring cross-cutting approaches over the human and computational sciences such as the following: (a) adversarial reasoning and behavioral game theory to predict attacker subjective utilities and decision likelihood distributions, (b) human factors of cyber tools to address human system integration challenges, estimation of defender cognitive states, and opportunities for automation, (c) dynamic simulations involving attacker, defender, and user models to enhance studies of cyber epidemiology and cyber hygiene, and (d) training effectiveness research and training scenarios to address human cyber-security performance, maturation of cyber-security skill sets, and effective decision-making. Models may be initially constructed at the group-level based on mean tendencies of each subject's subgroup, based on known statistics such as specific skill proficiencies, demographic characteristics, and cultural factors. For more precise and accurate predictions, cognitive models may be fine-tuned to each individual attacker, defender, or user profile, and updated over time (based on recorded behavior) via techniques such as model tracing and dynamic parameter fitting.

Numerosity estimates for attended and unattended items in visual search.

The goal of this research was to examine memories created for the number of items during a visual search task. Participants performed a visual search task for a target defined by a single feature (Experiment 1A), by a conjunction of features (Experiment 1B), or by a specific spatial configuration of features (Experiment 1C). On some trials following the search task, subjects were asked to recall the total number of items in the previous display. In all search types, participants underestimated the total number of items, but the severity of the underestimation varied depending on the efficiency of the search. In three follow-up studies (Experiments 2A, 2B, and 2C) using the same visual stimuli, the participants' only task was to estimate the number of items on each screen. Participants still underestimated the numerosity of the items, although the degree of underestimation was smaller than in the search tasks and did not depend on the type of visual stimuli. In Experiment 3, participants were asked to recall the number of items in a display only once. Subjects still displayed a tendency to underestimate, indicating that the underestimation effects seen in Experiments 1A-1C were not attributable to knowledge of the estimation task. The degree of underestimation depends on the efficiency of the search task, with more severe underestimation in efficient search tasks. This suggests that the lower attentional demands of very efficient searches leads to less encoding of numerosity of the distractor set.

Intentional control of event counting.

Event counting depends on simple, well-learned knowledge but is effortful and error-prone. In 6 experiments, the authors examined event-counting performance, testing a model that suggests that counting is controlled by minimal goal representations coordinated with perceptual events by temporal synchrony. In Experiment 1, they examined self-paced counting with or without delays that disrupted participants' preferred pacing. In subsequent experiments, participants counted computer-paced events occurring at rhythmic or varied intervals, reporting or verifying totals. Several results support the model: Participants counted rhythmic events more accurately, made undetected undercount errors when counting rhythmic events, and made false alarms to undercount or overcount probes presented at different times. These results suggest that intentions that guide fluent counting specify parameters deictically rather than semantically and that error monitoring is implicit.