The automatic and the expected self: separating self- and familiarity biases effects by manipulating stimulus probability