Positive and negative instructions in symbolic paired comparisons with the months of the year

Summary We investigated the semantic interpretation of positive (MORE THAN) and negative (LESS THAN) instructions in a symbolic paired comparison task with an overlearned and finite series of stimuli. French-speaking subjects were asked to indicate which member of month pairs occurred PLUS TOT (earlier), PLUS TARD (later), MOINS TOT (less early), or MOINS TARD (less late) in the year. In Experiment 1, the sign of the comparative instruction (MORE vs. LESS) was kept constant, while its attribute (EARLY vs. LATE) varied. In Experiment 2, the attribute of the instruction was kept constant, while its sign varied. The main results indicate that (1) negative instructions (with adverb LESS) yielded longer response latencies than positive ones (with adverb MORE); this difference was particularly salient on the first experimental session but, on subsequent sessions, it was reduced in Experiment 1 and neutralized in Experiment 2; (2) a symbolic-distance effect was observed on response latency and response accuracy with the four forms of the comparative instruction; (3) violations of monotonicity were noted in the symbolic-distance effect when June was used as a reference month in the analysis; (4) the serial-position effect had the form of a double-bowed curve with three stable anchor points around January, June, and December; and (5) the pattern of semantic-congruity effect with negative instructions was the exact reversal of that obtained with positive instructions; that is, if the month pairs occurred early in the year, response latencies were shorter with the instruction less late than with less early, whereas if the month pairs occurred late in the year response latencies were shorter with instruction "less early than with less late. The theoretical implications of these results for Banks' Semantic-coding model and Holyoak's Reference-point model are discussed.