There has recently been a review of the relationship between conscious and unconscious processing by van Gaal and others. (Citation below) They looked at an area of common misunderstanding and attempted to clarify it. Here are the highlights of their summarizing discussion:

… we have reviewed recent studies that have focused on the complexity and strength of unconscious information processing in relation to cognitive control (e.g., response inhibition, conflict resolution, and task-switching), the life-time of information maintenance (e.g., working memory, recognition memory) and the possibility to integrate multiple pieces of information across space and time. Unconscious information has been shown to affect various perceptual and high-level cognitive functions and the associated brain areas, including prefrontal cortex. In some cases, unconscious information has been observed to affect behavior and brain activity for relatively long periods of time. Overall, these recent results highlight the power of unconscious information processing, going beyond specific expectations formulated in traditional theoretical models of consciousness and the cognitive functions thought to require consciousness. … one can conclude that the potential function of consciousness might not be related to the initiation of cognitive control functions by specific stimuli that signal the need for increased control (e.g., stop-signals, task-switching cues). These cognitive control operations are probably triggered by a fast feedforward, and unconscious, early sweep of information processing that reaches even regions in the prefrontal cortex. This unconscious fast feedforward sweep can directly affect (the speed of) ongoing cognitive processes. … Although recent evidence has clearly pushed the boundaries regarding the duration of unconscious effects, the general observation is that unconscious events are much less able to elicit (long-term) future behavioral adaptations than conscious events (e.g., post-error slowing, conflict adaptation). Why might this be the case? Theoretical models of consciousness suggest that conscious awareness is related to long-lasting recurrent interactions between (distant) brain regions. This might enable the exchange of information between several spatially separated cognitive modules, which seems to break the automaticity of information processing. Awareness might be beneficial for enabling flexible and durable information processing strategies that are not directly triggered by a specific stimulus, for example when information has to be integrated across longer periods of time to bias information acquisition or signal the need for performance adjustments . Recently, Kunde et al. suggested that awareness might be dispensable when cognitive control is signaled explicitly (by specific control-eliciting stimuli) but not when it has to be inferred implicitly (by the context, or history of events).

van Gaal, S., de Lange, F., & Cohen, M. (2012). The role of consciousness in cognitive control and decision making Frontiers in Human Neuroscience, 6 DOI: 10.3389/fnhum.2012.00121