In the target article, Mitchell et al. argue that human associative learning requires participants to be consciously aware of contingencies between a conditioned stimulus (CS) and an unconditioned stimulus (US). This argument is strongly supported by extensive research on trace conditioning. Trace conditioning is a type of classical conditioning in which there is a temporal gap between the offset of the CS and the onset of the US. Trace conditioning is a well-established model of declarative learning that can be tested in both humans and nonhuman animals (Woodruff-Pak & Disterhoft Reference Woodruff-Pak and Disterhoft2008).

Awareness of the CS-US contingency has been suggested to be a prerequisite for successful learning of trace conditioning. This idea is supported by numerous studies on differential trace eyeblink conditioning, in which one of the stimuli (CS+) was always followed by an air puff to the eye (US), whereas another stimulus (CS−) was explicitly unpaired with the US. Clark and Squire (Reference Clark and Squire1998) found that only the subjects who were aware of the temporal relationships between the stimuli displayed differential responses to the CS+ and CS−. Their findings were replicated in a later study reporting that acquisition of trace eyeblink conditioning was significantly correlated with the awareness of stimulus contingencies (Knuttinen et al. Reference Knuttinen, Power, Preston and Disterhoft2001). Clark and Squire (Reference Clark and Squire1999) further showed that preventing the awareness of the contingency during conditioning disrupted differential trace conditioning, while providing knowledge about the relationship facilitated learning. Notably, the acquisition of the CS-US contingency and trace conditioning developed in a roughly parallel pattern (Manns et al. Reference Manns, Clark and Squire2000).

The necessity of awareness in trace conditioning has also been evident in fear conditioning studies (Carter et al. Reference Carter, Hofstotter, Tsuchiya and Koch2003; Knight et al. Reference Knight, Cheng, Smith, Stein and Helmstetter2004; Weike et al. Reference Weike, Schupp and Hamm2007). Using a differential trace fear conditioning paradigm, Weike et al. (Reference Weike, Schupp and Hamm2007) reported that the differential fear-potentiated startle and skin conductance responses (SCR) to the CS+ and CS− were observed during conditioning only when subjects were aware of the contingency between the CS+ and the US. In the study by Carter et al. (Reference Carter, Hofstotter, Tsuchiya and Koch2003), significant correlations between awareness and differential SCR responses to a CS+ and a CS− were present during extinction. Altogether, these findings indicate that awareness is required for acquisition and expression of fear in trace conditioning.

The importance of awareness in trace conditioning is reflected in its underlying neural mechanisms. Growing evidence suggests that trace conditioning is dependent on the hippocampus, a medial temporal lobe region which is widely believed to be critical for declarative memory (Clark et al. Reference Clark, Manns and Squire2002; Shors Reference Shors2004; Woodruff-Pak & Disterhoft Reference Woodruff-Pak and Disterhoft2008). Previous studies reported that amnesic patients who suffered from hippocampal atrophy failed to learn the CS-US contingency and, therefore, were impaired in trace eyeblink conditioning (Clark & Squire Reference Clark and Squire1998; McGlinchey-Berroth et al. Reference McGlinchey-Berroth, Carrillo, Gabrieli, Brawn and Disterhoft1997). The performance of the amnesic patients became worse as the trace interval between the CS and the US increased. Lesion studies in animals also show that aspirative or electrolytic lesions to the hippocampus disrupted trace eyeblink conditioning (Beylin et al. Reference Beylin, Gandhi, Wood, Talk, Matzel and Shors2001; Weiss et al. Reference Weiss, Bouwmeester, Power and Disterhoft1999; Solomon et al. Reference Solomon, Van der Schaaf, Thompson and Weisz1986). Notably, a recent single-unit study has shown that hippocampal CA1 neurons display highly accurate timed firing to the trace period. During the CS-alone testing session, CA1 neurons maximally fired in synchronization with trace interval used in the conditioning trials (McEchron et al. Reference McEchron, Tseng and Disterhoft2003). This timed firing was closely associated with behavioral responses to the CS. The findings suggest that hippocampal CA1 neurons are critically involved in encoding the trace interval, which is essential for the CS-US associations in trace conditioning.

With the advance of human brain imaging techniques, recent studies began to look directly into functional connections among awareness, hippocampus, and trace conditioning. A few fMRI studies have reported strong activations of the hippocampus during trace learning (Buchel et al. Reference Buchel, Dolan, Armony and Friston1999; Cheng et al. Reference Cheng, Disterhoft, Power, Ellis and Desmond2008; Knight et al. Reference Knight, Cheng, Smith, Stein and Helmstetter2004). The magnitude of hippocampal activation was closely associated with the accuracy of US prediction, which was a direct measurement of CS-US contingency awareness (Knight et al. Reference Knight, Cheng, Smith, Stein and Helmstetter2004). Besides the hippocampus, the activations of other brain regions, including middle frontal gyrus, that support attention and working memory were also associated with trace interval. One idea is that the hippocampus contributes to awareness by interacting with the neocortex (McIntosh et al. Reference McIntosh, Rajah and Lobaugh2003).

To sum up, trace conditioning requires an active process to form an internal representation of the contingency between relevant stimuli across a temporal gap. In this respect, the requirement of awareness for trace fear learning strongly supports the single-process propositional view of associative learning.