Since Fechner (1871) who confirmed in empirical studies that specific proportions, e.g. the “Golden Section”, are clearly preferred to others, composition principles in paintings have been a focus in aesthetic research. The “Compositional Pyramid” (Berger, 1963; Puttfarken, 2000) is another principle that is based upon the assumption that certain distributions of pictorial features (e.g. specific forms, lines, color and brightness) in a painting lead to the perception of structural balance. That is to say, effects of grouping on an automatic level of perception are influenced by deliberate artistic skill. For example Tyler (2006) showed by analyzing 170 portraits, each of a different artist from 15th to 19th century, that artists tend to
42 place one eye in the center of the painting. So there are rules of good composition that artists revert to when creating a picture, and influence the aesthetic processing of its beholder in specific ways. This stands in line with an artist’s aim to constitute meaning in an artwork by means of visual expression.
The power of different pictorial features to direct gaze in a picture is often referred to as saliency (Nothdurft, 2000) associated with bottom-up processes of visual perception. The saliency of different visual features in a picture can be computed in saliency-map-models predicting eye-movements to specific pictures (Itti, 2005). The tendency of participants to look at human features has been predicted by several saliency-map models. Showing natural scenes containing frontal shots of people, Itti and Koch (2001) found that faces were fixated within the first few fixations, whether subjects had to grade an image on interest value or search it for a specific non-face target. Cerf et al. (2008) found the tendency to look at human features to be stronger than the saliency-influence of other intrinsic image features such as color, organisation, intensity, flicker or motion.
In connection with saliency one might also mention the impact of detail on eye-movements on a painting. Using altered versions of 5 portraits of Rembrandt, Di Paola et al. (2013) showed that gaze can be directed by making specific areas of a picture especially salient using different ways of alternation, e.g. sharpening or highlighting. This is especially interesting, regarding specific schools in art like Mannerism, emerging in late Italian Renaissance, associated to emphasizing details by experimenting with proportion and perspective.
Locher et al. (1996) altered the structural balance of representational and abstract paintings and asked art-trained and untrained participants to indicate their center of balance. While representational art offers orientation on denotative details, the balance in abstract art is a matter of structure and visual regularities. The balance point was explained to participants as
43 the point where attention is naturally led. The results indicate a shift of perceived balance in original reprints and altered versions with high agreement between participants. Also, studies show that around 40% of the balance judgments are assigned to pictorial elements in the central area of a painting, for both trained and untrained participants (Locher et al., 1996; McManus, 1985). There are several explanations for this central bias. One theory, called the photographer bias, is trying to explain the central tendency in referring to picture-production: photographers tend to place objects or actors of interest near the center of their composition and enhance their focus and size relative to the background (Parkhurst & Niebur, 2003; Tatler, 2007). But the central bias can also be explained as a viewing strategy: Parkhurst et al. (2002) found out, that viewers reorient at a greater frequency to the center of a scene relative to other locations, if they expect highly salient or interesting objects there.
Visual weight and balance point are both synonyms for areas that get immediate or most attention (Arnheim, 1974; Locher et al., 1996; McManus & Kitson, 1995). Pictures look heavier on the right (Arnheim, 1974) and are inspected left to right, which is often explained with reading direction (Freimuth & Wapner, 1979), thus likely to be sensitive to culture (Chokon & De Agostini, 2000). Regarding emotional impact of composition, Heller (1994) showed that both adults and children place figures more to the left when painting sad pictures. Also, orientation changes the expressiveness of paintings. Balancing left-right effects, Bennett et al. (2010) found that the characters of animals in drawings by Thomas Bewick were rated less extreme when shown with orientation reverse to original. In sum, the presented studies indicate that composition influences meaning making on perceptual and emotional level by leading the eye and thus changing the distribution of attention to specific areas in a picture. A first selection of information in aesthetic processing is thus made due to automatic, bottom-up processes of perception.
44 Concerning aspects of good composition, studies also show that shapes can be more or less meaningful to the beholder, as proposed by gestalt psychology (e.g. Wertheimer, 1923; Köhler, 1929; Arnheim, 1949) and Gibson’s ecological approach to perception (1978). In the tradition of Berlyne (e.g. 1966), Shigeto and Nittono (2010) used a set of polygons to analyze the influence of complexity and meaningfulness on aesthetic experience. While complexity was expressed in the number of sides of the polygons, meaningfulness was rated prior to testing by asking 94 people to rate on a 7-point scale if the polygons looked like a namable object, but without naming the object. Complexity had an influence on how long participants looked at the polygons no matter how meaningful they were. Meaningful polygons were easier to recognize when rotated no matter how complex they were. Additionally, participants rated meaningful polygons as more interesting, in positive correlation with complexity. These findings show that it is not only structural properties of composition triggering certain emotions that entail attention (Chaterjee, 2003), but also cognitive associations (Marković, 2013) that sustain it.