Research Article

Classical Conditioning and Brain Systems: The Role of Awareness

See allHide authors and affiliations

Science  03 Apr 1998:
Vol. 280, Issue 5360, pp. 77-81
DOI: 10.1126/science.280.5360.77

Abstract

Classical conditioning of the eye-blink response, perhaps the best studied example of associative learning in vertebrates, is relatively automatic and reflexive, and with the standard procedure (simple delay conditioning), it is intact in animals with hippocampal lesions. In delay conditioning, a tone [the conditioned stimulus (CS)] is presented just before an air puff to the eye [the unconditioned stimulus (US)]. The US is then presented, and the two stimuli coterminate. In trace conditioning, a variant of the standard paradigm, a short interval (500 to 1000 ms) is interposed between the offset of the CS and the onset of the US. Animals with hippocampal lesions fail to acquire trace conditioning. Amnesic patients with damage to the hippocampal formation and normal volunteers were tested on two versions of delay conditioning and two versions of trace conditioning and then assessed for the extent to which they became aware of the temporal relationship between the CS and the US. Amnesic patients acquired delay conditioning at a normal rate but failed to acquire trace conditioning. For normal volunteers, awareness was unrelated to successful delay conditioning but was a prerequisite for successful trace conditioning. Trace conditioning is hippocampus dependent because, as in other tasks of declarative memory, conscious knowledge must be acquired across the training session. Trace conditioning may provide a means for studying awareness in nonhuman animals, in the context of current ideas about multiple memory systems and the function of the hippocampus.

Memory is composed of several different abilities that depend on different brain systems (1). A fundamental distinction is between the capacity for conscious recollection of facts and events (declarative or explicit memory) and various nondeclarative (implicit) forms of memory that are expressed in skills, habits, and simple forms of conditioning. This distinction is dramatically evident in amnesic patients, who have bilateral damage to the hippocampal formation or related midline diencephalic brain structures. These patients have severely impaired declarative memory and are profoundly forgetful. Yet these same patients have a fully intact capacity for nondeclarative memory (2). Indeed, a large body of literature involving both humans and experimental animals can now be understood by recognizing that memory tasks requiring declarative memory depend on the integrity of the hippocampal formation and related structures, whereas tasks requiring nondeclarative memory can be performed normally after damage to these structures and are supported by other brain systems. Declarative memory is what is meant by the term “memory” in ordinary language. It is involved in modeling the external world, and its contents can be brought to consciousness as a verbal proposition or as a mental image. By contrast, nondeclarative memory is expressed through performance without affording access to any conscious memory content or even awareness that memory is being used. This form of memory permits cumulative changes in perceptual and response systems and allows for the gradual development of new skills and habits.

A major puzzle about the distinction between conscious (hippocampus dependent) and nonconscious (hippocampus independent) forms of memory concerns classical conditioning. Classical conditioning, a phylogenetically early example of simple associative learning, has been studied extensively and would appear to be a quintessential example of nondeclarative memory (3). In perhaps the best studied classical conditioning paradigm, delay conditioning of the eye-blink response, a neutral conditioned stimulus (CS), such as a tone, is presented just before an air puff unconditioned stimulus (US). The US is then presented and the two stimuli coterminate (Fig. 1, A and B). Initially, an eye blink occurs reflexively in response to the US, but with repeated CS-US pairings a learned or conditioned response (CR) is elicited by the CS in advance of the US. The CR overlaps with the US such that the eye blink serves as an adaptive, defensive response to the air puff. Studies in the rabbit have shown that the cerebellum is essential for both the acquisition and retention of delay classical conditioning (4) and that no other forebrain structure, including the hippocampus, is required (5). Amnesic patients also exhibit intact acquisition and retention of the classically conditioned eye-blink response (6, 7). Thus, eye-blink conditioning appears to have the automatic, reflexive features that are characteristic of nondeclarative memory.

Figure 1

The temporal relationship between the CS and the US is shown for each conditioning procedure. During delay conditioning, the CS remained on until the 100-ms air puff US was presented and both stimuli coterminated. (A) The CS was presented for 700 ms before the US onset. (B) The CS was presented for 1250 ms before the US onset. (C andD) During trace conditioning, the CS was presented for 250 ms, and then a 500-ms trace interval (C) or a 1000-ms trace interval (D) intervened before presentation of the US. In each case, half of the trials involved presenting a second CS alone, without the US.

The puzzle concerns trace conditioning, a slightly different version of classical conditioning in which the CS is presented and terminated and then a short interval is imposed before the presentation of the US (8) (Fig. 1, C and D). The name comes from the fact that the CS must leave some trace in the nervous system for a CS-US association to be established. Trace conditioning requires the hippocampus (7, 9-11). Yet the trace interval (typically 1 s or less) is far too short to create any special difficulty for amnesic patients with hippocampal damage (12). Amnesic patients can easily hold onto information for many seconds. Accordingly, it has not been clear what aspect of trace conditioning requires the hippocampus or why trace conditioning might involve declarative memory.

We reasoned that trace conditioning might differ from delay conditioning by requiring knowledge of the CS-US relationship to build up and be remembered across many trials (13). To explore this possibility, we tested amnesic patients and control volunteers on both delay and trace conditioning, and we also assessed the knowledge that participants developed about the CS-US association.

Procedure. We tested 4 amnesic patients (14) and 48 normal volunteers (15) with two delay-conditioning procedures and two trace-conditioning procedures (Fig. 1). For delay conditioning, the CS was presented for 700 ms before the presentation of a 100-ms US (delay 700, n = 12), or the CS was presented for 1250 ms before US onset (delay 1250, n = 10). In both versions of delay conditioning, the CS and the US overlapped and coterminated. For trace conditioning, a CS was presented for 250 ms, and then a 500-ms trace interval (trace 500, n = 12) or a 1000-ms trace interval (trace 1000, n = 14) intervened before presentation of the US. The amnesic patients were given trace 1000 conditioning first and then 6 to 35 days later were given delay 1250 conditioning. All four conditioning paradigms used a differential conditioning procedure with two CSs in which one CS was consistently paired with the US (CS+) and a second CS was presented alone (CS). For half of the participants, the CS+ was a tone and the CS was white noise (static). For the other half of the participants, the CS+and the CS were reversed. In all four paradigms, training consisted of 60 CS+ and 60 CS trials, which were presented while participants viewed a silent movie. Differential conditioning was measured as the percentage of CRs to the CS+ minus the percentage of CRs to the CS(16).

Immediately after conditioning, participants took a true or false test that asked about aspects of the conditioning session, including how well they remembered the movie (Fig.2A) and how well they remembered the CS+, the CS, and the US and their responses to the CS+ and the CS (17). The critical questions were 17 additional items concerning the temporal relationships between the CS+, the CS, and the US. For example, true or false: I believe the air puff usually came immediately before the tone, I believe the tone usually came immediately before the static noise, I believe the static noise and air puff were always closely related in time, and I believe the tone predicted when the air puff would come. Participants responded to the test items in a fixed order and were not permitted to change earlier answers. In each of the four groups, participants who scored significantly above chance on the 17 critical questions [≥13 correct out of the 17 test items that asked about the temporal relationships between the stimuli (18)] were designated as aware of the relationships among the stimuli, and participants who did not score significantly above chance (≤12 test items correct) were designated as unaware (Fig. 2B). The acquisition of classical eye-blink conditioning (19) was then compared for the aware and unaware groups on each version of the task.

Figure 2

(A) Number of correct responses to 10 true or false questions about the content of the silent movie that participants watched during the conditioning procedure. AMN, four amnesic patients. Each of the four other groups consisted of 10 to 14 control participants given delay eye-blink conditioning or trace eye-blink conditioning (see Fig. 1). The AMN group performed no better than chance and more poorly than each of the other groups (allP values < 0.001). (B) Number of correct responses to 17 true or false questions about the temporal relationships between the CS+, the CS, and the US. The black bars are the scores of control participants who were aware of the CS-US relationship, the white bars are the scores of control participants who were unaware of the CS-US relationship, and the hatched bars are the scores for the four amnesic patients. The number of participants in each group is the same as in (A). Error bars show the SEM.

Experimental results. Knowledge of the stimulus contingencies was not related to performance on either version of delay conditioning, but it was a crucial factor in both versions of trace conditioning (Fig. 3). Specifically, normal volunteers acquired delay conditioning whether they were knowledgeable about the CS-US associations or not. For trace conditioning, only those individuals who developed knowledge of the CS-US associations successfully acquired the task. Finally, the amnesic patients, none of whom became aware of the CS-US associations (20), were unable to acquire trace conditioning, although they acquired delay conditioning at a normal rate (Fig. 3). The number of correct responses on the true or false test (out of 17 items that asked about the CS+, CS, US relationship) was not correlated with the percentage of differential responding for either version of delay conditioning (delay 700, r = −0.10, P > 0.5; delay 1250, r = 0.16, P > 0.5) but was significantly correlated with the percentage of differential responding for both versions of trace conditioning (trace 500, r = 0.74, P < 0.01; trace 1000, r = 0.69, P < 0.01).

Figure 3

Performance during classical conditioning of the eye-blink response by amnesic patients (AMN) and four groups of normal volunteers. The data are presented as the percentage of differential conditioned eye-blink responses for each block of 20 trials (percentage of CRs to the CS+ minus percentage of CRs to the CS). Each 20-trial block included 10 CS+trials in which a tone (or white noise) occurred together with an air puff to the eye (US) and 10 CS trials in which white noise (or a tone) occurred but the US did not. The temporal relationship between the CS+ and the US is shown at the top of each panel. The participants in each condition were classified as aware or unaware according to their performance on a 17-item, true or false test that asked about the relationship between the CS+, the CS, and the US. Participants scoring >2 SD above the chance score of 8.5 correct (that is, 13 correct) were considered aware. Only the aware groups and not the AMN group or the unaware groups acquired differential trace conditioning (C and D). All the groups acquired delay conditioning (A and B), and, unlike trace conditioning, awareness was not a factor in the learning of differential delay conditioning. The SEMs ranged from 0.03 to 0.08. (A) ▪, aware (n = 3); •, unaware (n = 9). (B) ▪, aware (n = 7); •, unaware (n = 3); ○, AMN (n = 4). (C) ▪, aware (n = 5); •, unaware (n = 7). (D) ▪, aware (n = 7); •, unaware (n = 7); ○, AMN (n = 4).

For trace 1000, the failure of the amnesic and unaware groups to demonstrate differential conditioning was due to the failure to acquire CRs to the CS+. For trace 500, unaware participants failed to demonstrate differential conditioning because they did not discriminate between the CS+ and the CS. That is, they exhibited CRs to both the CS+ and the CS (21).

We also addressed the nature of the relationship between awareness and conditioning. Did awareness occur as a result of conditioning, or did conditioning occur because participants became aware of the CS-US associations? In both versions of delay conditioning, participants acquired differential conditioning even if they did not become aware of the CS-US associations. This finding shows that successful conditioning does not guarantee awareness. To examine this issue more directly in the case of trace conditioning, we tested two new groups of participants. With the first group (five men and three women, mean age = 70), we thoroughly explained the temporal relationships between the CS+, the CS, and the US before trace 1000 conditioning. All participants exhibited differential conditioning. After conditioning, the group obtained a score of 16.0 correct out of 17 on the test items that asked about the temporal relationships between the stimuli. This group also exhibited significantly improved conditioning compared with the 14 control participants in the trace 1000 group (block six scores, P < 0.05; Fig. 3D) (22). The second group (five men and three women, mean age = 63) was given trace 1000 conditioning while concurrently performing an attention-demanding task (detecting strings of three odd digits in a running sequence). This group obtained 9.0 correct out of the 17 test items that asked about the stimuli, and they did not exhibit differential conditioning (block six scores = −0.05%, significantly poorer than the 14 control participants in Fig. 3D; P < 0.05). This finding strengthens the notion that awareness is a prerequisite for successful trace conditioning.

Implications. The results appear to resolve the puzzle of why trace conditioning depends on the integrity of the hippocampus. Like other tasks of declarative memory that are impaired after hippocampal lesions, trace conditioning requires the acquisition and retention of conscious knowledge across a considerable time span (in this case, the 30-min conditioning session). Specifically, individuals must acquire and retain knowledge of the task structure if conditioning is to be successful. In earlier work, the limited ability that amnesic patients had for acquiring tasks of declarative memory correlated with their ability to verbalize the principles underlying the tasks (23). Trace conditioning may require declarative knowledge because the trace interval between the CS and the US makes it difficult to process the CS-US relationship in an automatic, reflexive way. This more complex condition likely requires the neocortex to represent the temporal relationships between the stimuli (24) and would require the hippocampus and related structures to work conjointly with the neocortex to establish a usable representation that can persist as memory.

Trace conditioning is dependent on the cerebellum as well as the hippocampus (25). Thus, even though trace conditioning differs from delay conditioning in its requirement for declarative memory, it resembles delay conditioning in that a nondeclarative learning circuit in the cerebellum is required for the generation of the conditioned response. Thus, in the case of trace conditioning, it appears that a representation of the CS-US relationship, dependent on the hippocampus and neocortex, can then be used by the cerebellum to support conditioning. Task awareness may develop whenever the hippocampus and neocortex are engaged during learning.

The concept of conscious knowledge is not readily applied to experimental animals. Nevertheless, an implication of the present findings is that learning and memory tasks, including trace conditioning, which are failed by animals with hippocampal lesions, are tasks about which intact animals must acquire declarative knowledge. Characteristics that have been helpful in extending the concept of declarative memory to nonhuman animals include its flexibility and the ability to use it inferentially in novel situations (26). The conjoint operation of the hippocampal system and the neocortex may be the critical element that confers awareness about knowledge that has been acquired (27, 28).

The finding that trace conditioning requires subjects to become aware of the temporal relationships among the stimuli explains why trace conditioning is declarative and hippocampus dependent, and it brings classical conditioning, the best studied of all learning paradigms, into register with current understanding of the memory systems of the brain.

REFERENCES AND NOTES

View Abstract

Navigate This Article