Reducing negative stimulus valence does not attenuate the return of fear: two counterconditioning experiments