Abstract: There is a debate between the multiple systems opinion and the single system opinion in probabilistic category learning, and the experiments of researchers holding different opinions have adopted different ways of presenting the positions of the cues respectively. So using the classical weather prediction task, the current study manipulated the ways in which the positions of cues were presented to explore the influence that this had on probabilistic category learning.
This study included two experiments. Experiment 1 investigated the learning systems by fixing and randomizing the positions of all the cues. Experiment 2 investigated the strategy through fixing the positions of the singleton cues.
The results showed that when the positions of all cues were held constant, probabilistic category learning was an explicit learning process. However, when the positions of all cues were random, it was an implicit learning process. And when only the positions of the singleton cues were held constant, it was also an implicit learning process.
These results indicate that the different ways of presenting the positions of the cues affect the competition between the explicit learning and implicit learning systems, which supports the multi-system opinion. Moreover, the main strategy in probabilistic category learning may be the multi-cue strategy rather than the singleton strategy.

Key words: probabilistic category learning, weather prediction, explicit learning, implicit learning, rolling regression