EVOLUCIÓN DE LOS ASTROCITOMAS DESTACANDOSE EL VALOR DE LOS DATOS HISTOLÓGICOS.

It is, of course, impossible to directly observe the cognitive processes of an individual performing a task. Recording one’s eye movements, or observing one’s physical behaviours, gestures, and interactions with task artefacts both offer data sets that can be interpreted in terms of underlying cognition (e.g. Rayner, 1997; Underwood, Chapman, Bowden, & Crundall, 2002). However, to collect these forms of data whilst an individual is engaged in the driving task can be costly (in time and resources), distract from the task at hand (and therefore change cognition), and provide data sets that are difficult to interpret (in terms of cognitive mechanisms and structures). One such method that attempts to circumvent these issues is verbal protocol analysis (Ericsson & Simon, 1980, 1993).

The technique has two primary variants; concurrent think-aloud and retrospective think-aloud. Concurrent think-aloud requires an individual to verbalise their thoughts concurrently with task performance; in retrospective think-aloud the participant provides verbal reports after the task has finished. Each has its advantages and disadvantages, and each may be more suited to certain environments or domains (e.g. Banks, Stanton, & Harvey, 2014b; Russo, Johnson, & Stephens, 1989; van den Haak, De Jong, & Jan Schellens, 2003).

Ericsson and Simon (1980) speak of different types, or levels, of verbalisations that may be produced. They do so in terms of the potential for intervening, recoding processes to occur in the time between information leaving the central processor (in working memory; e.g. Atkinson & Shiffrin, 1968; Baddeley & Hitch, 1974) and the production of the verbalisation. Three levels of verbalisation are specifically described. Level one represents a direct reproduction of information from working memory; this undergoes no intermediate processing. Level two involves the recoding of non-verbal internal representations into verbal code. This does require an additional degree of translation but crucially does not alter a person’s cognitive processes (Ericsson & Simon, 1993); the individual is not expected to explain their thoughts, nor is

attention directed or manipulated by the researcher. The third level does just this; it requires either the explanation (rather than simply reporting) of thoughts, a scanning or filtering of thoughts (for particular referents in the environment or task), or a necessity to verbalise aspects to which the participant would not normally attend. Ericsson and Simon explain this in terms of the difference between explaining and thinking aloud:

“When subjects are asked for explanations, the reports cannot be generated without extending the information and relations heeded […] Hence, thinking aloud, as distinguished from explanation, will not change the structure and course of the task processes” (1980, p. 226).

It is, therefore, verbalisations at levels one and two in which we are interested when seeking to understand basic cognitive processes and structures. Hence the approach used in this study was to use concurrent think-aloud without directing the participants’ attention to any particular objects or events, requiring the participants to simply report their thoughts as opposed to explaining them. As aforementioned both concurrent and retrospective reporting have advantages and disadvantages, and both may be suited to particular tasks or environments (e.g. Russo et al., 1989). It has been argued that retrospective reporting suffers from problems of non-veridicality, i.e., the lack of correspondence between verbal reports and cognitive processes, particularly in tasks of long durations (e.g. van Gog, Kester, Nievelstein, Giesbers, & Paas, 2009). The driving sessions in the study presented in this chapter lasted for just under 20 minutes (comparable to the average journey time of 23.7 minutes in the UK; Department for Transport, 2014). Retrospective verbal reports of tasks of this length are not only subject to issues of omissions and post-task fabrications (van Gog et al., 2009), but can be biased towards positive aspects (Swann, Griffin, & Predmore, 1987) and effective actions (Kuusela & Paul, 2000). Moreover, Nisbett and Wilson (1977) demonstrated that participants providing retrospective verbalisations are no more accurate in the identification of the determinants of behaviours than are observers. I therefore argue the concurrent think-aloud procedure to be a more appropriate technique in the context of the current study.

There are a number of examples in the literature of the concurrent think- aloud procedure as applied to the driving domain. One early example comes from Hughes and Cole (1986) in their investigation of the foci of attention when driving. In this study, however, one could argue that the verbal protocol analysis procedure (as argued for by Ericsson & Simon, 1980, 1993) was not strictly followed, insofar as participants were specifically asked to report that which drew their attention (hence resulting in the potential for level three verbalisations). Indeed, when talking of the intermediate scanning or filtering of thought processes, Ericsson and Simon (1980) themselves talk of the driving domain, making reference to Soliday and Allen’s work (see Ericsson & Simon, 1980, p. 219) in which participants were asked to report all perceived traffic hazards. The act of directing attention requires the participant to first scan the environment and then categorise that which they see before verbalising their thoughts, thereby representing level three verbalisations. Therefore these cannot be said to accurately reflect underlying cognitive processes.

More recently, Walker et al. (2001a) reported an on-road investigation in which concurrent verbal reports contributed to an understanding of the role of feedback in driver cognition, and Lansdown (2002) used concurrent verbal reporting in a driving simulator study to investigate individual differences in drivers, highlighting differences between novices and experts. It is possible to find additional such studies more recently still; for example, Walker and colleagues furnish us with two more uses of the concurrent think aloud procedure in the driving domain. In Walker et al. (2011) the technique was again applied in an on-road setting, with the aim of investigating the differences in the ways car drivers and motorcyclists interpret road situations. Results highlighted some ‘critical incompatibilities’ (Walker et al., 2011, p. 878) between the two road user groups. In Walker et al. (2013) the technique was used for the analysis of situation awareness, also in an on-road environment; however, the participants here were asked to ‘explain their actions’ (Walker et al., 2013, p. 21), hence may have suffered from the presence of level three verbalisations (following Ericsson and Simon’s arguments).

Further examples of the on-road use of concurrent verbal reporting can be found in work by Young and colleagues (2015; 2013). In both studies, the first

of which investigated distraction-induced driver error (Young et al., 2013), the second investigating attention at rail-road crossings (Young et al., 2015), verbal protocol analysis was used as an additional source of information, enriching the data set built up from interviews, videos, and the output of the instrumented vehicle. Banks et al. (2014b) also used the concurrent think aloud procedure as an additional source of information in a driving simulator study. In their study, an exploratory analysis of verbal reports was used in conjunction with quantitative simulator data, with the verbal reports adding value to the understanding of the thought processes that underlie behavioural outcomes. Finally, Pampel et al. (2015) provide, to my knowledge, the only instance of a driving study to use the concurrent think aloud procedure for the specific investigation of eco-driving. The authors compared the verbal reports of people when asked to drive ‘normally’, ‘safely’, and ‘efficiently’, drawing on mental model research in discussing the simulator-based study’s results. The study’s findings led the authors to argue for the existence of mental models specific to eco-driving; models that are not employed when asked to drive ‘normally’. The authors drew attention to the existence of misconceptions concerning speed and travel time. Whereas Pampel et al. (2015) were interested in the differences exhibited by participants when driving for different guiding purposes (i.e., normal, safe, or efficient), this chapter is interested simply in the potential of verbal reports to help build an understanding of what distinguishes more efficient drivers from less efficient drivers, without any guidance of behaviour or driving style.

Specifically, it deals with the possible differences between those who are more or less fuel-efficient when driving ‘normally’, i.e., driving without additional instruction or guidance. The question is, therefore; do drivers who display more fuel-efficient driving styles (as evidenced by quantitative vehicle data) also show differences in their underlying cognitive structures and processes (as revealed in their concurrent verbal reports)? This question is, of course, highly exploratory in nature, and does not invite specific hypotheses; I simply seek to identify common strategies held by those who exhibit more efficient driving styles, and investigate the ways in which they might differ from those who drive in a less economical manner.