Emotional reactivity to visual scenes affects both how we attend to them and how they
are remembered, but it is not clear how these attention and memory effects are related.
Weapon-focus theories (e.g., Loftus, 1979) suggest that attention is restricted to
emotion-provoking parts of scenes, and that such restriction of attention affects the
specificity of the memory that is stored. I directly tested whether “weapon-focus-like”
restriction of attention predicts subsequent visually-specific memory for emotional
scenes by recording eye movements while participants viewed relatively emotional and
relatively non-emotional slides during initial encoding. Even though visually-specific
memory was equivalent for all types of scenes, different patterns of eye movements
predicted subsequent memory for emotional and non-emotional scenes. For emotional
scenes only, visually-specific memory was predicted when eye movements were
restricted to emotional parts of the scenes during encoding. For non-emotional scenes,
visually-specific memory was predicted when more fixations of shorter duration were
made, and attention was relatively broadened across the scene during encoding.
Experiments 2 and 3 tested whether these patterns of eye movements reflect local or
global processing of scenes, but across both experiments, there was no evidence that
local and global processing influenced subsequent memory effects. The subsequent
memory effects from Experiment 1 were replicated in Experiment 2, but in Experiment 3, a relative broadening of attention – more fixations of shorter duration – predicted
subsequent memory for all scenes, a pattern that had only been observed for nonemotional
scenes previously. Experiment 4 was conducted to test whether this occurred
because emotional reactions to emotional scenes were reduced by having participants simply view each scene, which reduces emotional responses when compared to cases
where valence and arousal judgments are made as they were in previous experiments.
Experiment 4 replicated the subsequent memory effects from Experiment 3,
demonstrating important boundary conditions on the subsequent memory effects
established in the first two experiments. These results suggest that qualitatively distinct
memory representations may be stored for emotional and non-emotional scenes, but
both representations are capable of supporting visually-specific memory.