Attitudes can be formed and changed with evaluative conditioning. In evaluative conditioning (EC),
an initially neutral conditioned stimulus (CS; e.g., an unfamiliar brand name) is paired with a valent
unconditioned stimulus (US; e.g., a pleasant picture). After repeated pairings, the evaluation of the CS
is changed in line with the US’s valence (i.e., the EC effect). It is debated whether EC reflects
(semantic) memory of the pairings, or is due to a separate learning process that is independent of a
participant’s awareness of the pairings. The former account predicts EC effects to be associated with
pairing memory, whereas the latter predicts EC in the absence of pairing memory. Here we studied EC
effects for different subjective memory states - remembering, knowing, intuition/gut-feeling, and
random guessing. Across three studies, EC effects were associated with above-chance pairing
memory. We consistently found EC to be strongest for 'remember' states and absent for 'guess' states.
We also found weaker EC effects for 'intuition/gut-feeling' states (which were associated with less
accurate pairing memory). These results support an association between EC and pairing memory. Our
findings also point to the possibility of EC effects occurring with weak memories that may not be
subjectively experienced as such. Practical implications include the necessity for control strategies to
address these weaker memory states.