Relating Personality and Physiological Measurements to Task Performance Quality

Relating Personality and Physiological Measurements to Task Performance Quality Fiemke Both, Mark Hoogendoorn, Rianne van Lambalgen Vrije Universiteit Amsterdam, Department of Artificial Intelligence De Boelelaan 1081a, 1081 HV Amsterdam, The Netherelands Email: {fboth, mhoogen, rm.van.lambalgen}@few.vu.nl Rogier Oorburg, Michael de Vos Force Vision Lab, Barbara Strozzilaan 362a, 1083 HN Amsterdam, The Netherlands Email: {rogier, michael}@forcevisionlab.nl Abstract Human performance can degrade as a result of an increasing cognitive workload. Especially in domains in which good task performance is crucial these effects are unwanted, and hence, need to be avoided. In Psychological literature studies have been presented which relate a variety of measures to the workload experienced by humans. However, these experiments have not been conducted with more complex tasks. In this paper, the aim is to perform a variety of measures to see whether combinations thereof can be used to predict performance of humans under tasks with varying complexity. A dedicated experiment has been conducted with 31 subjects, of which the results have been analyzed from both a statistical as well as a temporal perspective. Keywords: human workload, human personality, physiological measures. performance, Introduction Working under high pressure can result in negative effects on the functional state of a human (see e.g. Hancock 1995, Hockey 1997). As a result of this negative effect, the performance of the task at hand can be negatively influenced. In critical domains such degradation of performance can lead to severe consequences that might be highly undesired (Wickens, 2002) Therefore it is very useful to be able to obtain more precise knowledge about measurements that can be used to make more accurate predictions concerning the performance upon tasks (and the potential degradation thereof). In the literature, various measurements have been proposed to measure the cognitive workload experienced by humans. Roscoe (1992) for example, shows that the heart rate of a human is a good indicator of cognitive workload (if no physical effort is performed). Also subjective measurements have been put forward, such as the NASA- TLX questionnaire (Hart and Staveland, 1988). In (Rose et al., 2002) the relation between the so-called Big Five personality factors upon vigilance performance and workload has been studied. However, the tasks studied do not concern complex tasks. In this paper, the aim is to perform a set of measurements to see whether the performance of a task can be predicted for complex tasks as well using (a combination of) these measurements. Hereby, an experiment has been conducted whereby participants had to perform a cognitive task with varied complexity. The measurements performed consisted of: (1) personality characteristics (using the NEO-PI-R and NEO-FFI test, cf. Costa and McCrae, 1992); (2) cognitive abilities (using several tests); (3) performance quality using an objective measure; (4) the complexity of the task; (5) the NASA-TLX as a measure of perceived workload, and (6) the heart rate during the experiment. For the data obtained, the prediction is that performance will depend on personality characteristics, as well as on heart rate and the complexity of the task. Furthermore, the subjective performance as measured with the NASA-TLX is predicted to be dependent on objective performance and personality. For the data analysis, not only a statistical analysis is performed, but also more complex temporal properties have been analyzed. For example trends over time that influence the performance quality and perceived workload will be analyzed. In order to enable such a temporal analysis, tools from the domain of Artificial Intelligence are used (cf. Bosse et al., 2008). Using these analysis methods, the relationship between personal characteristics (like personality, cognitive abilities and expertise) and performance is investigated for complex tasks, as well as between task characteristics (e.g. the complexity of a task) and performance. These kinds of relations can be used to give adequate support, for example if a correlation would be found between heart rate and performance, a high heart rate would indicate that another human should be signaled to support the performance of the task. This paper is organized as follows. In Section 2 the experimental setup is described. Section 3 describes the statistical non-temporal analysis whereas Section 4 descries the temporal analysis. Finally, Section 5 is a discussion. Experimental Setup Participants In this study 31 people participated (18 males, 13 females), of which 25 students from the Vrije Universiteit.