Search

With Artificial Intelligence taking on the roles and responsibilities of full-fledged team members there is a dire need to assess the ability of these ‘Human-Machine’ teams and compare them to Human-Human teams. Humans are influenced by the social context of the teaming environment and will adapt their behaviour accordingly if they are to compete or collaborate with another person. How the behaviour of Humans in Human-Machine teams is influenced by their artificial co-actor and the teaming environment remains unclear as do the effects of this behaviour on performance. Furthermore, the additional strain of teamwork (such as anticipating and integrating co-actor behaviour) can come at a cost to the team’s performance or rather hinder performance efficiency. Global performance measures alone cannot capture the cost of teamwork nor explore the behaviours that lead to the success or failure of teams. We utilised a novel design to explore the differences in the performance efficiency and behaviour of Human-Human teams and Human-Machine teams. Performance in both Team types indicated teamwork came at a cost in efficiency, but the cost was far less for Human-Human teams under a Collaborative teaming environment. When exploring the behavioural patterns, we found Collaboration for Human-Human teams led to a decrease in redundant behaviour and that behaviour was a significant predictor of performance efficiency. Additionally, the measure used to capture redundant behaviour appeared to be a valid indication of dynamic team behaviour. The joint analysis of performance efficiency and behavioural patterns provided a better insight into the way human team members work alongside human and non-human players. Such analysis can further inform theories regarding the ways human players interpret the actions of other team members - humans and AI - and incorporate this knowledge to drive their own actions.

Despite the ubiquitous nature of Evidence Accumulation Models in cognitive and experimental psychology there has been a comparatively limited uptake of such techniques in the applied literature. While quantifying latent cognitive processing properties has huge potential for applied domains such as adaptive work systems, accumulator models often fall short in practical applications. Two primary reasons for these shortcomings are the complexities and time needed for the application of cognitive models, and the failure of current models to capture systematic trial-to-trial variability in parameters. In this manuscript we develop a novel, trial-varying extension of the Shifted Wald model to address these concerns. By leveraging conjugate properties of the Wald distribution we derive analytic solutions for threshold and drift parameters which can be updated instantaneously with new data. The resulting model allows the quantification systematic variation in latent cognitive parameters across trials and we demonstrate the utility of such analyses through simulations and an exemplar application to an existing data set. The analytic nature of our solutions opens the door for real-world application, significantly extending the reach of computational models of behavioral responses.

In the modern world, there are important tasks that have become too complex for a single unaided individual to manage. Some safety-critical tasks are conducted by teams to improve task performance and minimize risk of error. These teams have traditionally consisted of human operators, yet nowadays AI and machine systems are incorporated into team environments to improve performance and capacity. We used a computerized task, modeled after a classic arcade game, to investigate the performance of human-machine and human-human teams. We manipulated the group conditions between team members; sometimes they were incentivised to collaborate, sometimes compete, and sometimes to work separately. We evaluated players’ performance in the main task (game play) and also measured the cognitive workload they experienced. We compared workload and game performance between different team types (human-human vs. human-machine) and different group conditions (competitive, collaborate, independent). Adapting workload capacity analysis to human-machine teams, we found performance under both team types and all group conditions suffered a performance efficiency cost. However, we observed a reduced cost in collaborative over competitive teams within human-human pairings but this effect was diminished when playing with a machine partner. The implications of workload capacity analysis as a powerful tool for human-machine team performance measurement are discussed.

The word superiority effect refers to the phenomenon that people have better recognition of letters presented within words as compared to recognition of isolated letters. Although many previous research on how the spatial relations between letters in words affect the perceptual processing through the inversion paradigm, a significant amount of effort goes into setting the default inter-letter spacing when designing new fonts. Our current research examines the effect of manipulating letter spacing on the processing efficiency, as a measure of the word superiority effect. First, we tested multiple different words instead of fixed word stimuli to show that measures of efficiency can be generalized; second, we disrupted default inter-letter spacing by increasing, decreasing, and randomizing letter spacing to explore the extent to which the efficiency was sustained with the assessment functions. Our results indicate that participants are limited capacity only in the extreme spacing scenario. Additionally, the principle component (PC) analysis shows that highest PC values occur at normal spacing with degradation with increased disruption—spreading or narrowing. These results appear to confirm the configural nature of perceptual processing with normally-spaced words between identifiable tracking and kerning boundaries, and agree well with the ideas about optimal spacing by type designers and typographers implicit in general notion of "rhythmic spacing''. This work is also notable in that we demonstrate the use of assessment functions as a standardized tool for assessing the capacity benefits and efficiency of configural processing.

Complex-valued periodic sequences, u, constructed by Goran Bjorck, are analyzed with regard to the behavior of their discrete periodic narrow-band ambiguity functions Ap(u). The Bjorck sequences, which are defined on ℤ/pℤ for p>2 prime, are unimodular and have zero autocorrelation on (ℤ/pℤ)∖{0}. These two properties give rise to the acronym, CAZAC, to refer to constant amplitude zero autocorrelation sequences. The bound proven is \(|A_{p}(u)| \leq2/\sqrt{p} + 4/p\) outside of (0,0), and this is of optimal magnitude given the constraint that u is a CAZAC sequence. The proof requires the full power of Weil’s exponential sum bound, which, in turn, is a consequence of his proof of the Riemann hypothesis for finite fields. Such bounds are not only of mathematical interest, but they have direct applications as sequences in communications and radar, as well as when the sequences are used as coefficients of phase-coded waveforms.

Given a discrete sample of event locations, we wish to produce a probability density that models the relative probability of events occurring in a spatial domain. Standard density estimation techniques do not incorporate priors informed by spatial data. Such methods can result in assigning significant positive probability to locations where events cannot realistically occur. In particular, when modelling residential burglaries, standard density estimation can predict residential burglaries occurring where there are no residences. Incorporating the spatial data can inform the valid region for the density. When modelling very few events, additional priors can help to correctly fill in the gaps. Learning and enforcing correlation between spatial data and event data can yield better estimates from fewer events. We propose a non-local version of maximum penalized likelihood estimation based on the H 1 Sobolev seminorm regularizer that computes non-local weights from spatial data to obtain more spatially accurate density estimates. We evaluate this method in application to a residential burglary dataset from San Fernando Valley with the non-local weights informed by housing data or a satellite image.

Direct simulation Monte Carlo was used to study ion and neutral transport and reaction in a low-gas-pressure highplasma-density inductively coupled reactor with chlorine (electronegative) chemistry. Electron density and temperature were computed by a self-consistent continuum plasma code and were used as input to the direct simulation Monte Carlo code. Simulation results were compared with experimental data taken in a Gaseous Electronics Conference reference cell