Abstract
In two interval forced choice (2IFC) tasks it has been known since Fechner's time (1860) that the stimulus in the second interval is perceived as being stronger for weight discrimination. Not only is there a response bias, the thresholds (JNDs) are systematically lower in one interval than in the other. Recent studies by Nachmias (Vision Res. 2006) implicate the role of a memorized reference as being responsible for this asymmetry whereby when the test is in the first interval it perturbs the magnitude of the reference.
Our recent experiments were designed to test some of the hypotheses for explaining this Time Order Error (TOE). Our new intervention was that for each 2IFC adaptive staircase trial on a contrast discrimination task we introduce a small contrast increment or decrement (jitter) to both intervals of a given trial. If observers based their judgment on the difference of the two intervals the sign of the jitter would be irrelevant. If observers compared the 2nd interval to a weighted average of the 1st interval and a memorized reference, they would have lower JND values when the test was in the 2nd interval. We interleaved four 2IFC separate staircases for the two intervals and the two signs of jitter.
The results for two subjects on a Gabor contrast discrimination task (30% reference) were nearly identical. There was a strong bias effect whereby interval 2 seemed stronger than interval 1 about 70-75% of the time. The JND was significantly lower in interval 2, similar to that reported by Nachmias on very different tasks. The jitter parameter demonstrated that both observers paid more attention to the 2nd interval, by about 13+/-6%. Although our jitter manipulation explains only a small part of the TOE, it provides a powerful tool for assessing the decision process in 2IFC tasks.