In the eyes of the beholder: How experts and novices interpret dynamic stimuli

https://doi.org/10.1016/j.learninstruc.2009.02.019Get rights and content

Abstract

Tasks with a complex, dynamic visual component require not only the acquisition of conceptual/procedural but also of perceptual/attentional skills. This study examined expertise differences in perceiving and interpreting complex, dynamic visual stimuli on a performance and on a process level, including perceptual and conceptual strategies. Performance, eye movement, and verbal report data were obtained from seven experts and 14 novices. Results show that experts compared to novices attend more to relevant aspects of the stimulus, use more heterogeneous task approaches, and use knowledge-based shortcuts. Implications for instructional design for the acquisition of perceptual skills are discussed.

Introduction

Many instructional materials directly or indirectly convey expert knowledge to learners (Feldon, 2007). For instance, a very effective instructional technique is the use of worked examples, in which novices are shown a worked-out expert solution procedure (Sweller, Van Merriënboer, & Paas, 1998). However, worked examples have so far mainly been used to convey conceptual and procedural aspects of expertise. In many domains, expert performance also comprises perceptual/attentional skills, that is, the ability to perceive the relevant out of irrelevant information in complex, highly visual stimuli and to draw inferences based upon the perceived information (e.g., X-rays in medical diagnosis see Lesgold et al., 1988; weather maps in meteorology see Canham & Hegarty, 2010).

An important question is, if these perceptual/attentional aspects of expertise, that is top-down processing of perceptual stimuli, could also be conveyed to novices to facilitate skill acquisition. Since providing information at a conceptual level has been shown to be effective for the acquisition of conceptual/procedural skills, it might be necessary to grant novices more direct access to the perceptual/attentional processes underlying experts’ performance; for example, by guiding the novices’ attention to critical perceptual information during the study of worked examples based on evidence regarding experts’ perceptual processes, such as eye movements (Van Gog, Jarodzka, Scheiter, Gerjets, & Paas, 2009; for process-oriented worked examples see Van Gog, Paas, & Van Merriënboer, 2008). However, as will be discussed in the next section, not much is known about how experts allocate their attention during task performance, and how their attention allocation differs from novices, especially in domains that involve complex, dynamic visual stimuli. Therefore, the present study examined expertise in the perceptual/attentional and conceptual processes involved in the interpretation of complex dynamic visual stimuli in the domain of fish locomotion.

Many studies have addressed the issue of how experts perceive and interpret visual stimuli by using eye tracking methodology. In particular, these studies provide information on attention allocation through eye movement analyses. Gazes on relevant information have higher densities than on irrelevant information. Haider and Frensch (1999) stated in their information-reduction hypothesis that with increasing expertise people learn to distinguish between relevant and irrelevant information and therefore concentrate on processing mostly relevant information (see also Canham & Hegarty, 2010). Using a letter string task in which the location of relevant information was varied, Haider and Frensch (1999) corroborated this hypothesis with eye movement data. In the domain of art, Antes and Kristjanson (1991) found that experts (i.e., artists) compared to novices (i.e., non-artists) had higher fixation densities on important aspects of the paintings. Charness, Reingold, Pomplun, and Stampe (2001) also found in their studies of expertise effects in chess performance that experts had a greater proportion of fixations on relevant rather than on irrelevant areas.1

The above studies, however, used static visual stimuli; few eye tracking studies have taken place using dynamic visual stimuli (e.g., for air traffic control see Ellis, 1986) to identify differences in visual attention in general, and none has yet investigated attention to visually complex and dynamic stimuli that contain relevant and irrelevant information. One exception to the use of dynamic material is the study by Moreno, Reina, Luis, and Sabido (2002), in which novice and expert gymnastic coaches inspected videos of gymnastic techniques and indicated errors in performance. They found that experts had longer and fewer fixations than novices, and attributed this to the fact that experts attended more to informative (i.e., relevant) areas and ignored uninformative (i.e., irrelevant) ones. However, this assumption was not directly tested in the study. Another exception is the study by Lowe (1999) on the interpretation of weather maps, where novices mentioned more often irrelevant but perceptually salient features after the inspection of the dynamic weather maps, suggesting that they attended more to these features. However, this assumption was not directly tested with eye tracking.

So far, eye tracking has provided interesting insights into how experts differ from novices using single basic eye tracking indicators (e.g., number or duration of fixations) when processing tasks with a high visual component. Little research has been done to identify expertise effects in perceptual strategies used, that is, complex patterns of eye movements when processing dynamic visual stimuli. There are at least two open questions associated with this issue.

First, one may ask whether experts’ perceptual strategies are characterized by an optimization of the strategies that a novice would use or whether experts act within their domains in a qualitatively different way than novices do. In the case of conceptual processing it has been shown that experts’ highly integrated knowledge structures enable them to use shortcuts during task processing (e.g., medical diagnosis based on textual descriptions of cases: Boshuizen & Schmidt, 1992), which results in very different strategies for different expertise levels. In contrast, in case of perceptual processing it has been found that experts process and report visual case information more elaborately than novices (e.g., for diagnosing X-ray pictures see Lesgold, 1984), which rather seems to favor the idea that experts optimize strategies that are used imperfectly by novices.

Second, it is unclear whether experts’ perceptual strategies are the same for all experts or differ among experts (cf. Medin et al., 2006). For example, Medin, Lynch, Coley, and Atran (1997) showed that, while some experts were very similar in their approaches to the task (namely, parks maintenance personnel vs. scientific taxonomists in categorizing trees) other experts differed in their approach to the same task (namely, landscapers vs. scientific taxonomists). Moreover, Medin et al. (2006) showed in another study that even in well-structured domains, like the categorization of freshwater fish, expertise did not lead to common conceptualizations. The authors concluded that even if the outcome of a categorization process is similar across experts the underlying processes are not necessarily so.

The above conclusion, however, might be due to the characteristics of biological taxonomies in which one has to deal with a diversity of the features of the various species. These taxonomies are often invented, that is, they are conceptual schemas not inherent to a domain; hence, there may be multiple categories for the same species depending on the focus of the taxonomy (e.g., morphology, genetics, etc.). Furthermore, categorizing species is based on multiple features, which do not need to be considered in a hierarchical order. Experts may differ in the features that they emphasize in order to achieve a categorization depending on their prior experiences, that is, on their learning history. For instance, an expert with lots of outdoor experience will potentially focus on other features (e.g., those easily observable in a natural setting) than an expert who mostly deals with formalin preparations and textbooks (e.g., features requiring a close inspection). Hence, these differences in learning history, which will be reflected in the organization and accessibility of knowledge, may affect how experts attend to features (for endogenous attentional control see Corbetta and Shulman, 2002, Posner, 1980) yielding diverse perceptual strategies for experts. On the other hand, novices’ attention may be mainly guided by exogenous features, for instance, by salience (Lowe, 1999). Consequently, their perceptual strategies might be rather homogeneous compared to those of experts.

Therefore, two issues are very important when studying expertise differences: (a) namely the specialization of the experts under investigation; and (b) the nature of the task (Medin et al., 1997, Medin et al., 2006). When investigating perceptual strategies, the task to-be-performed is even more important because eye movements strongly depend on it (Yarbus, 1967).

In sum, many tasks require the use of perceptual skills. However, instructional materials seldom teach these skills directly. To design instructional material that is suited to convey perceptual skills, knowledge on expertise differences in the analysis of complex, dynamic visual stimuli is necessary. Although some studies dealing with complex, static visual stimuli exist, research on interpreting complex, dynamic visual stimuli is still rare. The aim of the present study was to investigate expertise differences as a prerequisite for designing effective instructional material in the domain of fish locomotion.

In the present study expertise differences were investigated for the description of locomotion patterns of fish swimming forward. This task was chosen for two reasons. First, this task is dynamic and has a highly visual component, thus, perceptual strategies are very likely to play a crucial role in task processing. Second, it is a good example of the topic of biodiversity, which is a core topic in biology and part of the curriculum at school and university, because fishes are the most diverse vertebrates; besides a large diversity of forms, colors, and habitats, they are also very diverse in terms of their locomotion patterns (Videler, 1993).

To describe a locomotion pattern, the following guidelines should be applied (Lindsey, 1978). First, it has to be decided which part of the body is used to produce propulsion. This can be either the body itself or the fins. Second, it has to be decided how this part moves. This can be either in an undulating (i.e., wavelike) or an oscillating (i.e., paddlelike) way. These decisions form the basis of what will be called a locomotion description strategy here, which is taught to university students of biology when they receive training in marine zoology. Application of this strategy, would lead the observer to attend only to those parts of the fish body that might be crucial for the fish's locomotion (i.e., the fins and the body) and ignore the parts that are irrelevant to the locomotion (i.e., eyes or colorful patches).

Both novices and experts have to rely on the locomotion description strategy when classifying locomotion of unfamiliar fish. However, with familiar fish, experts may be able to automatically retrieve knowledge on the specific locomotion pattern associated with this particular fish species from memory. In such a case a species classification strategy will be used, and other features of the fish are potentially relevant. Specifically, some species can be easily recognized due to salient static features (e.g., a specific colorful pattern on the fish's body is a characteristic of a particular fish species). Experts can use these features to classify a fish, and upon activating that particular schema, the knowledge on its locomotion pattern is automatically activated. For novices, this type of knowledge-based shortcut is not available, so they need to rely on the locomotion description strategy.

This study investigated how different levels of expertise in biology would be reflected in differences in task performance (i.e., correct description of the locomotion) as well as in different strategies. These differences were investigated by means of eye tracking and cued retrospective verbal reports (cf. Van Gog, Paas, Van Merriënboer, & Witte, 2005). Using these methods in conjunction was expected, first, to provide insights into the perceptual and conceptual processing of complex visual dynamic stimuli and, second, to provide an example of how eye tracking data on dynamic stimulus material can be analyzed in detail. In addition, findings obtained from this study were intended to inform the instructional design of process-oriented worked examples that teach perceptual skills by guiding students’ attention (Van Gog, 2009).

It was hypothesized (Hypothesis 1) that experts would perform more accurately and faster than novices on a locomotion description task (reflected in higher correctness rates and shorter mean viewing times of the stimulus). The respective test, however, mostly served as a manipulation check concerning our assignment of individuals to different levels of expertise.

More important, based on prior findings with static, complex visual stimuli (Antes and Kristjanson, 1991, Charness et al., 2001, Haider and Frensch, 1999) it was hypothesized (Hypothesis 2) that also in this dynamic domain the process data of experts would show that they attend more to relevant information than novices, who would rather attend to perceptually salient, but potentially conceptually irrelevant information (cf. Lowe, 1999). Whether experts attend to features that are relevant for either locomotion description or species classification was explored by means of analyzing the gaze duration on these features. It was predicted that experts would attend more features relevant to the species than novices (Hypothesis 3).

Moreover, whether expertise yields either more diversity or more homogeneity in terms of the perceptual strategies used was investigated by analyzing experts’ and novices’ gaze sequences. It was assumed that novices’ perceptual strategies will be guided by the salience of single features and, thus, possibly leading to a more homogeneous gaze pattern. On the other hand, experts’ perceptual strategies were assumed to be controlled in an endogenous way. Hence, experts were expected to be characterized by rather heterogeneous gaze patterns depending on their individual learning history that would lead to different processing strategies (Hypothesis 4).

Finally, it was expected that experts would verbalize less information than novices due to schema automation and thus use fewer words in their description of how they accomplished that task (Ericsson & Simon, 1980). In line with Boshuizen and Schmidt (1992) their verbalizations were expected to contain more encapsulating technical terms (Hypothesis 5).

Section snippets

Participants and design

Participants in the study were 21 individuals (M = 26.57 years, SD = 5.98; 10 females and 11 males) with two different levels of expertise. Of them seven were experts, that is, professors, PhDs, or advanced PhD students, with a mean age of 31.43 years (SD = 8.54). The novices were 14 biology students of the University of Tuebingen, Germany, with a mean age of 24.14 years (SD = 1.51), who had basic knowledge of fish anatomy, terminology and locomotion pattern, but had very little if any experience with

Results

An alpha level of .05 was used for the statistical tests reported.

Discussion

The present study aimed at identifying expertise effects in perceiving and interpreting complex, dynamic visual stimuli, both at an outcome and a process level. It was hypothesized that experts would perform more accurately and faster than novices (Hypothesis 1) and that the process data of experts would show that they attended more to relevant information than novices, who would attend more to irrelevant information (Hypothesis 2). Furthermore, Hypothesis 3 addressed the issue of which

Acknowledgements

This work is part of a larger research project on “Resource-adaptive design of visualizations for supporting the comprehension of complex dynamics in the Natural Sciences” funded by the Leibniz–Gemeinschaft.

During the realization of this work Tamara Van Gog was supported by a Rubicon grant from the Netherlands Organization for Scientific Research (NWO; Grant nr. 446-07-001).

References (28)

  • N. Charness et al.

    The perceptual aspect of skilled performance in chess: evidence from eye movements

    Memory and Cognition

    (2001)
  • M. Corbetta et al.

    Control of goal-directed and stimulus-driven attention in the brain

    Nature Review Neuroscience

    (2002)
  • S.R. Ellis

    Statistical dependency in visual scanning

    Human Factors

    (1986)
  • K.A. Ericsson et al.

    Expert and exceptional performance: evidence of maximal adaptation to task constraints

    Annual Reviews in Psychology

    (1996)
  • Cited by (307)

    View all citing articles on Scopus
    View full text