The human observer effortlessly perceives both the spatial layout of the environment and the shapes of significant objects. Apart from these geometrical aspects, the observer perceives the qualities of objects, that is to say aspects of their chemical constitution, physical properties, and so forth. For instance, one easily classifies objects as made of wood, metal, paper or cloth, ... being polished, rough or hairy, ... being dry, wet or greasy, ... This project aims at the development of novel tools to probe human perception of material properties.

We start from the assumption that physical descriptions of the stimulus domain are available. Possible methods are: measurement of discrimination thresholds in a parameterised continuum of stimuli, direct scaling methods in which the subject rates stimuli on a subjective scale, indirect scaling methods in which the subject compares stimuli with respect to a perceptual attribute and methods of adjustment in which the observer is asked to adjust parameter values such that a certain perceptual state is achieved.

Methods of adjustment are convenient because they allow one to obtain much data in a relatively short period and they appear as very "natural" tasks to the observers: In the case of the perception of material properties the problem is much the same as that faced by the painter. For instance, how should one change a coloured area such that it looks "shiny", "wet", or "metallic"? We are often in a position to predict which "knobs to turn" in order to arrive at a desired effect.

We have identified various aspects of the proximal stimulus that are likely candidates as "cues" for the perception of material attributes. These can be parametrically varied in a controlled manner thus enabling psychophysics. We plan to work on aspects of texture, BRDF (Bidirectional Reflection Distribution Function) and surface roughness.