Fig. 1.

Hypothesis and methods. (a) In ACX, we tested three possibilities for CS encoding: (1) CS encoding represents their physical properties and is thus similar to encoding of never-reinforced NS; this may be the same for simple and complex sounds, or different (as illustrated here). (2) CS encoding reflects threat predictions for simple and complex sounds alike, i.e. the difference between CS and NS is the same for simple and complex sounds. (3) CS encoding represents threat only for complex but not simple sounds. An incidental button press task controlled attention and sensory discrimination for all sound pairs. (b) Frequency-spectrograms of complex and simple sounds. We compared differential fear conditioning of frequency-modulated sweeps (complex) with single sine tones (simple), (c) Fear conditioning paradigm during fMRI. Participants were presented with reinforced and non-reinforced sounds in alternating order. They were instructed about the context (reinforced or non-reinforced) but not about the CS-US associations.