Running experiments on designers' perceptual judgements

Co-Proposer	Prof Claudia Eckert, Open University
Software	Java or C++ or Smalltalk or another object oriented language with good GUI facilities
Covers	GUI building, application development, possibly linking to statistics package
Skills Required	Programming, preferably interest in design or psychology
Challenge	Conceptual ??? Technical ??? Programming ???

Brief Description

An important aspect of design is the ability of skilled and experienced designers to make rapid perceptual judgements of the correctness and quality of aspects of their own and others' designs. This is especially significant in industries where visual appearance is important, both for technical correctness and aesthetic appeal. Fashion design and knitwear design are important examples where this is clearly the case; but perceptual evaluation plays an important role in engineering and architecture. But perceptual evaluation of the quality and correctness of designs has been very little studied; designers take it for granted, and non-designers aren't aware of its importance. (We found no discussions or experimental studies to cite when referring to this point in a research paper.)

The objective of the project is to build a computer system that runs psychological experiments to study designers' and non-designers' perceptual judgements. We are interested in this because we want to understand the role of perceptual skills in designers' expertise and how this is developed, as well as how designers' memories work for all the garments they have seen or can imagine. A good software system would make running well-designed experiments a lot easier.

Some of the experimental paradigms we would like to use the system to try are (1) The experimental subjects are asked to look at two curves and (a) say whether they are the same length or one or the other is longer and (b) give a confidence value for their judgements. (2) The experimental subjects are asked to look at cutting patterns for garment pieces, or curves representing part of a cutting pattern, and assess their correctness. (3) The experimental subjects are asked to rate the similarities of different garments (or other types of designs). We may think of other experimental tasks we want to use the program for, or you may suggest alternatives.

The program should show either single bitmaps (loaded in from a library), or pairs of bitmaps, along with the questions the subjects should answer and instructions for entering their answers. The program should be designed so that it can step through a (changeable) sequence of self contained modules, for each experimental trial and for screen presentations of initial instructions, explanations of what the experiments are about, and so on.

Extensions

The program would be more flexible and useful if it could display successive screens (within a single experimental trial) either in a precisely-timed sequence, or at intervals that are controlled by the user's input but are precisely measured.

The program should ideally contain a clean mechanism for defining the both contents of individual experimental trials and the order of trials, to make it easier to run experiments where the order is systematically varied.

The program would gain from doing its own statistical analysis so that results can be shown immediately to experimental subjects, either by including its own statistical routines or by calling a statistics package.

The program would be more useful if it could be distributed as a stand-alone application that can be used by subjects in their own homes or offices without the experimenter needing to be present.

Back to