Previous experiments suggest that the direction of SNARC is fixed for numbers based on a mental number line. This theory is challenged as our experiments show that for all symbols representing semantic quantities, including numbers, the direction of SNARC is dependent on instruction such that faster right hand and upward responses were always associated with the highest measure corresponding to the instruction, and faster and downward responses were always associated with the lowest measure corresponding to the instruction. In Experiment 1, an instruction-dependent SNARC effect was obtained for animal size, and Experiment 2 demonstrated that SNARC could be obtained for a more abstract form of magnitude such as animal intelligence. In Experiments 3, 4, and 5, participants were required to pick the warmer or colder digit from positive and negative pairs representing relatively high or low temperatures. Presentation of pairs and responses were horizontal in Experiment 3 and vertical in Experiments 4 and 5. An instruction-dependent SNARC effect occurred such that responses to the "coldest" and "warmest" pairs were always fastest to the right, or upward depending on the instruction, but left or downward responses were faster for pairs least compatible with the instruction. These results show that numbers based on a specific semantic property may be processed differently than nonspecific numbers. Additionally, a thermometer appeared on the screen for half of the trials in Experiments 4 and 5, which elicited a stronger SNARC effect demonstrating the importance of the semantic properties of stimuli. In Experiment 6, the instruction dependent pattern of SNARC continued using the speed of runners represented by positive and negative numbers. These results suggest that the direction of SNARC is caused by two additive factors including the absolute value of numbers, as well as the semantic quantity represented by numbers defined by the instruction. In the absence of one factor, the other factor dominates. SNARC in these experiments is best explained by the assignment of codes to the poles of stimulus and response dimensions and the compatibility of those codes. When response and stimulus codes are compatible responses are fast, but slower when they are incompatible.
展开▼