Evaluating architectural interiors with terms of everyday language

Predicting or evaluating the quality of spatial situations is a major unsolved problem in architecture, because of the lack of generally accepted objective methods. Whereas psychophysics or cognitive science use exact methods and concepts, their common measurands and categories unfortunately do little to objectify the character of architectural interiors. Judgements in everyday language are not clearly divided into different perceptional, cognitive or emotional categories and seem at a first glance subjective and fuzzy. However, they reflect the human perception and awareness directly and naturally, and, systematic patterns provided, they may lead to generalizable predications. This study was motivated by two main goals: First an exemplary statistical investigation about conciseness and intersubjectivity of colloquial judgements in general, and second an evaluation of assumed relations between judgements and measurable features of the scene. In a brainstorming session with 24 participants most common and subjectively important categories to characterize interiors were collected. Subsequently, pictures of 15 different interiors were rated in 12 of those categories by 42 subjects in two groups (architects and laypersons). The experiment used a novel internet based questionnaire technique derived from the semantic differential. Each category was represented by a pair of oppositional adjectives and evaluated on a nine step Likert-like scale. Conciseness and intersubjectivity differed from category to category, but averaged ratings showed clear correlations: Interrelations within evaluation categories could be demonstrated as well as correlations to basic image features of the evaluated interiors (color, brightness, saturation, amount of edges). In accordance with the assumptions, for example, warmth of the scene correlated with coziness (r2=0.82) as well as with red rgb value (r2=0.66). Furthermore, group specific differences were clearly apparent: architects responded more consistently and tended to prefer more austere designs. Rating spatial situations using terms of everyday language provided qualitatively and quantitatively meaningful results. Especially systematic comparisons to image features seem to be a viable method for getting further insights into the underlying signification of judgements. Extending this data basis appears to be a promising way towards predicting and quantifying the character and ambience of rooms by their physical features