As described before, “freedom of gaze” or the ability to control ones attention is an important attribute of art experts, because it enables the beholder to focus on meaningful information for the interpretation of a picture (Kesner, 2006; Zembylas, 2003; Mackworth & Bruner, 1970). This is why in experimental aesthetics eye-tracking is a method often used to compare experts and laymen in art. It is assumed that eye-movements are linked to attention and thus reflect cognitive processing (eg. Buswell, 1935; Massaro et al., 2012) not only bottom-up but also top-down, e.g. through instruction, training or expertise (Yarbus, 1967; Antes & Kristjanson,
47 1991; Jarodzka et al., 2010). Top-down driven gaze is associated with image aspects of higher order such as style, semantics and context (Cerf et al., 2008). Areas that are especially looked at are referred to as relevant. To predict eye-movements according to relevance, areas of interest are defined in a picture by making assumptions in close connection to task.
A pioneer in eye-tracking, Yarbus (1967), found that participants, rather than randomly scanning a picture, frequently fixate specific parts of a paining that carry meaning. With the help of an art educator, Kristjanson and Antes (1989) defined centers of interest in nine representational paintings as areas that provide important information for meaning making. 15Artists and 15 non-artists, who viewed these paintings for 20s, had a greater density and duration of fixations on centers of interest compared to other areas. Using the same data Antes and Kristjanson (1991) discriminated artists from non-artists analyzing fixation density on non-informative areas of the paintings. Concerning the group of artists it was interesting, that areas with little information for meaning making were more frequently looked at when paintings were unfamiliar. When they knew a painting, they were looking less at non- informative areas and thus were more effective in analyzing the painting for meaning making. Nodine et al. (1993) manipulated the balance of six paintings ranging from representational to abstract art and showed the original and manipulated version for 12s each in randomized order to seven participants trained and seven untrained in art. Fixation durations on the paintings were clustered and aggregated according to specific areas in the paintings. Art- trained participants made longer fixations on the original, balanced versions than on the altered, unbalanced versions of the paintings. For non-artists it was the opposite. Interpreting short fixations as a sign for global exploration and long fixations as a sign for local
48 exploration, the authors assume that experts are sensitive to composition and use the structure in original paintings as “skeleton” that leads their attention and helps them analyzing the painting intensively on areas that are informative for meaning making. Laymen on the other hand cannot make use of a painting’s structure for meaning making, so their gaze is lost in information.
Presenting a set of 35 paintings belonging to five categories in a continuum from representational to abstract art in two sessions for 10 and 30s, Phiko et al. (2011) recorded the eye-movements of 20 art historians and 20 students of no art related studies. Defining the sum of all saccade durations as length of the scanpath the eye of a participant made over a painting, they found that with rising abstraction of a painting the length of the scanpath and the number of fixations increased, while the mean duration of fixations decreased for all participants. Thus the gaze patterns of both experts and laymen for abstract art are characterized by short fixations with global scanning and for representational art by longer fixations with local scanning.
Zangemeister et al. (1995) presented five paintings of different genres to six participants without art training, four participants interested in art and four artists. In contrast to Phiko et al. (2011) here eye movements showed that artists scanned the paintings more globally, especially abstract art, while non-artists’ scanpaths of representational and abstract paintings showed no significant difference. But it has to be said that the measurement of scanpath for Zangemeister et al. (1995) was defined differently, namely the ratio of global and local saccades, split by their amplitude lower or above 1,6°. Also, the number of stimuli as well as participants is quite small in this study and may have led to an overestimation of found differences.
49 Nodine et al. (1993) also found that artists looked longer at areas in the background and figures changed by balance-manipulation, while non-artists look longer at foreground and central figures. Having a closer look at the fixations on human features for an inspection time of 30s, assuming that this time is sufficient for coming to an aesthetic judgment Phiko et al. (2011) found that laymen looked longer at human features than experts. Other eye-tracking studies (Vogt, 1999; Vogt & Magnussen, 2007) confirm this finding that untrained viewers prefer looking at human features and objects, while trained participants spent more time on scanning structural and abstract features. Yarbus already observed this greater freedom of experts from boundness to human features in 1967. Vogt and Magnussen (2007) found that artists’ gaze is stronger influenced by task instructions. When asked to memorize paintings (12 representational, 4 abstract) experts made local fixations on representative paintings and global fixations on abstract ones. In sum these results point to that experts can adjust their gazing behavior to the information-offer of a painting, traceable in composition and style.
4.4 Verbal Data as Source for Investigating Aesthetic Processing on Higher Levels of