Empirical Assessment of Detection and Classification Performance with a Two-Dimensional Color Code.((R.D. Beer¹, K.A. Jameson¹, J.L. Kaiwi²)) University of California, San Diego¹; United States Navy NCCOSC RDTE DIV²

PurposeThe accurate visual representation of information, as in the representation of data from a brain imaging device, is of paramount importance. Perceptual irregularities are known to occur when humans examine and interpret visual stimuli, and in the visual representation of data this must be taken into account. This research investigates the visual coding of information using a two-dimensional color code, and differences in performance when the code is constructed by two different methods. Methods. Two separate two-dimensional color codes were constructed, both coding signal strength (by the brightness dimension) and signal identity (by the chromatic dimension). One code was based on equal luminance steps, another based on existing psychophysical and perceptual data. Three experiments were conducted with 5 knowledgeable and 18 naive subjects using 376-trial signal detection tasks. A subset of subjects was assessed using a monochrome code to establish a baseline performance. Results. Detection efficiency using a chromatic information code was comparable to performance with a monochromatic code. Detection and identification sensitivity for both color codes was close to predicted values. Efficiency was comparable between the two color codes used, except in one condition. Conclusions. A two-dimensional color code (using brightness and hue) is effective at conveying information, and the two coding schemes tested produced roughly equivalent performance. However, one tested coding scheme is algorithmically produced, and therefore can be used easily to generate reliable color codes for a wide range of situations. Supported by U.C.S.D Academic Senate Grant #RV141G, National Research Council Associateship 94/95, and U.C.S.D. Undergraduate Scholastic Grant #43S96.