Abstract：The valence and arousal of emotional stimuli are thought to be the main factors affecting emotionally enhanced memory (EEM). A large body of neuroimaging evidence has indicated that there is a distinction between arousal dependent amygdala?hippocampal network implicated in automatic encoding processes and valence dependent prefrontal cortex (PFC)?hippocampal network implicated in controlled encoding processes. In a study, Kensinger and Corkin (2004) found that distinct cognitive and neural processes contributed to the enhancement of emotional memory for arousing versus valenced, nonarousing information. However, positive emotional stimuli were not included in their experiments. Accumulating evidence has shown that negative and positive stimuli can involve different cognitive and neural processes, it is therefore that, more evidence, especially about the memory for positive emotional stimuli, is needed to make stronger claims about this hypothesis. Previous studies have proved that attentional resources have a greater impact on the controlled than the automatic processes. If controlled encoding processes were responsible for the enhancement effect for negative nonarousing items but were less important for that for arousing items, the enhancement for the negative nonarousing items should be disproportionately reduced by the divided-attention as compared to that for the arousing items. In this study, three experiments were conducted using the learning?recognition and divided attention (DA) paradigm to explore the cognitive processes contributing to the EEM effects under the condition of valence (negative and positive) or arousal was anchored. All participants performed encoding tasks for emotional and neutral words in two different attentional conditions, FA (full or nondivided attention) or DA (divided attention). After the entire learning phase was completed, they performed a rehearsal prevention task and then a recognition test. Recognition scores were computed by subtracting the false alarm rate from the hit rate (Hits-False alarms). The results suggest that in FA condition, there exist both arousal-dependent and valence-dependent EEM effects. What is particularly important here is that there is no significant difference in the recognition scores for the negative arousing words between the two attentional conditions, but the recognition scores for the positive arousing words, negative and positive nonarousing words were significantly higher in FA condition than in DA condition. In other words, a distraction task during encoding has less influence on memory for negative arousing stimuli, but has an adverse impact on memory for both negative and positive nonarousing stimuli and positive arousing stimuli. Therefore, the EEM effect relied on valence is associated with controlled processes, whereas the EEM effect relied on arousal is mediated by the valence of the stimuli, which is associated with automatic processes for negative stimuli, but controlled processes for positive stimuli. The present findings implied that the EEM effects dependent on valence and arousal do not perfectly correspond with controlled and automatic processing.