Current Biology, Vol. 13, R65–R67, January 21, 2003, ©2003 Elsevier Science Ltd. All rights reserved.
Prefrontal Cortex: Procedural Sequence Learning and Awareness E.M. Robertson and A. Pascual-Leone
Activation of the prefrontal cortex has been linked to awareness during sequence-learning tasks. A recent study, however, finds activation of the prefrontal cortex during such tasks regardless of awareness. So what is the neurophysiological basis of awareness, and what is the role of the prefrontal cortex in sequence learning?
Being aware of learning new information, whether as facts, events or skills, is a familiar experience to many of us. Imagine sitting at a piano trying to learn the sequence of keys to press with a given speed and force to play a simple sonata. You know what you wish to accomplish, you practice and eventually you are able to play the music. The acquisition of a skill through practice is termed ‘procedural learning’. In this case you are aware of your effort to learn. What is perhaps less often appreciated is that procedural learning can also occur without awareness of the knowledge being acquired. You listen to your daughter playing the piano, while you wonder about her getting older and her dedication to the music. Later you find yourself tapping out a sequence of finger movements and your wife notices that you are ‘playing’ your daughter’s piano exercise. You did not realize it, but despite being distracted by other thoughts you have acquired the sequence of finger movements necessary to play the music. At times, such learning might occur even though you are unable to recall the circumstances of learning. You may, for example, not even remember watching your daughter play, but you may still have learnt the sequence of her finger movements. A novel sequence of finger movements can be learnt either with — ‘explicit’ learning — or without — ‘implicit’ learning — awareness of the underlying pattern. Experimentally, the serial reaction time task [1,2] has been used extensively to explore the influence of awareness on the recruitment of neural circuits during procedural learning (Figure 1). Functional imaging studies have contrasted the brain circuits activated during implicit and explicit learning of this task and found very little overlap between the two. One study [3] found only the thalamus activated during both implicit and explicit sequence learning, despite both conditions requiring roughly the same behaviour. Consistent with the view that awareness is the principle explanation for the differential recruitment of Laboratory for Magnetic Brain Stimulation, Behavioral Neurology Unit, Beth Israel Deaconess Medical Center, Harvard Medical School, 330 Brookline Ave, KS 454, Boston Massachusetts 02215, USA. Email:
[email protected]
PII S0960-9822(02)01431-8
Dispatch
cortical areas during sequence learning, the prefrontal cortex has been found activated exclusively during explicit sequence learning [4]. Recruitment of evolutionarily more advanced areas of the cortex such as the prefrontal cortex, with presumably greater computational sophistication, may explain the dramatic increase in the rate of learning that occurs with awareness. A recent functional imaging study [5] has challenged these arguments, however, by showing that the prefrontal cortex is recruited regardless of awareness for the underlying sequence. In earlier experiments [3,6], awareness of an underlying sequence was allowed to develop with prolonged practice. In this design, awareness develops only when great skill has been achieved, so that skill and practice cannot be fully dissociated from awareness. Other experiments have used a secondary distracter task to reduce the propensity of developing awareness for a sequence [7–9]. Unfortunately, this design distorts the very nature of the task, so that differentiating between effects of awareness and the distracter becomes impossible. An elegant solution to these problems is to have the same sequence of finger movements acquired almost simultaneously under both implicit and explicit conditions [5]. In the approach of Willingham et al. [5] a series of red dots guides the acquisition of what subjects know to be a sequence of finger movements, creating an explicit-overt condition (Figure 1). When the dots change colour from red to black, subjects are told that the dots follow no pattern. But this is only true for the random condition; at other times, unbeknownst to the subjects, the black dots can also follow a pattern. At times, the sequence originally denoted by red dots is re-played disguised as black dots. In this explicit-covert condition, the sole difference between responding to the black and the red dots is a subject’s awareness of following a sequence. Alternatively, an entirely novel sequence can be followed by the black dots in an implicit condition. This task is similar to those used in previous studies to investigate the acquisition of skill without awareness of an underlying sequence. Using functional imaging, the brain areas associated with the performance of each of these conditions were observed. Those specifically related to skill acquisition were pinpointed by comparison to the random condition [5]. Each of the three con -trasts showed a very similar pattern of activation. The implication is that sequence learning recruits essentially the same neuronal machinery, regardless of awareness for the skill being acquired. Specifically, the results show that prefrontal cortex is activated in explicit, as well as in implicit, procedural sequence learning. The contemporary challenge is not so much to determine whether the prefrontal cortex is making a contribution to procedural learning, but rather to
Dispatch R66
Figure 1. The serial reaction time task. A single ‘dot’ appears on a computer screen at one of four possible positions Stimulus within a horizontal array. Each position on the screen corresponds to a button on a 1 response box. When a dot appears on the screen a subject selects the appropriate button as quickly and accurately as possible. The time between presentation of the stimulus on the screen and selection 2 of the response is the ‘reaction time’. Upon selecting the correct response, the dot disappears to be replaced after a 3 1 2 4 fixed delay by a further dot. The subjects Response are told that the dots can be red or black. Subjects are informed that the red dots 12 follow a sequence — this is the explicit overt condition — but black dots appear randomly, no longer following a pattern. In fact this is not always the case. After Response Response changing colour to black, the dots can Delay Time Time play out one of three possible conditions. In the random condition, the black dots Current Biology are truly presented without any underlying sequence. In the explicit covert condition the black dots are presented following the identical sequence to that played out by the red dots in the explicit overt condition. Finally, in the implicit condition the black dots follow an entirely novel sequence. Comparing across these conditions in a functional imaging experiment Willingham et al. [5] were able to control for the effects of practice and performance and dissociate these from awareness. This study demonstrates that the same neural circuit is recruited during procedural sequence learning regardless of a subject’s awareness for the skill being acquired.
discern the nature of this contribution. When appropriate controls are introduced, the prefrontal cortex is activated during implicit sequence learning and shows even greater activation during explicit learning [5]. These differences might reflect distinct roles being undertaken by the prefrontal cortex during implicit and explicit sequence learning. The prefrontal cortex may play only a subsidiary role during implicit sequence learning, perhaps not being directly involved in learning at all, but instead ushering in awareness for the underlying sequence [5]. This suggestion receives some support from recent observations [10] which show that the dorsolateral prefrontal cortex has a role in searching for a recurring pattern amongst random events. It might be argued that confronted with a series of events, humans always try to find an organizing rule or pattern, hence engaging the prefrontal cortex to keep track of preceding events using working memory [11]. If so, the contribution of the prefrontal cortex may be the same in explicit and implicit sequence learning. Studies which explored procedural learning in the setting of disruption or damage to the prefrontal cortex have demonstrated its critical role in implicit sequence learning. Patients with ischaemic damage to the prefrontal cortex fail to show implicit sequence learning of tasks involving the hand contralateral to the affected hemisphere [12]. This result confirmed earlier work in which high frequency transcranial magnetic stimulation (TMS) was used to temporarily disrupt function within the dorsolateral prefrontal cortex in normal subjects: this was found to impair sequence learning with the hand contralateral to the targeted hemisphere [13–15]. TMS of other brain sites, such as the supplementary motor area, failed to reproduce this deficit.
Awareness, however, is not the only possible explanation for the recruitment of the prefrontal cortex during procedural sequence learning. Different types of sensory cue can guide the acquisition of a sequence. These cues may determine the pattern of recruitment across the prefrontal cortex. In its basic form, the serial reaction time task [1,2] (Figure 1) uses the spatial position of a stimulus to guide a response. An arbitrary mapping between a colour and a response allows colour cues, appearing always in the same position, to guide sequence learning. A recent study [14] found that disruption of the dorsolateral prefrontal cortex prevented the implicit acquisition of a sequence guided by spatial cues, while learning a sequence guided by colour cues was unhindered. The usually enhanced skill observed when both cues simultaneously guide the acquisition of the same sequence was reduced to a level commensurate with only a single cue being available to guide learning. Recruitment of specific parts of the prefrontal cortex, as suggested by the domain-specific theory applied in working memory, may reflect the cue guiding procedural learning [16] and be independent of awareness. The question of what may be the organising principle for the recruitment of cortical areas, particularly the prefrontal cortex, during sequence learning remains unanswered. What seems certain, thanks to the new work of Willingham et al. [5], is that awareness no longer offers a unifying explanation. Instead, awareness appears to be an emergent property of the activation of a given brain circuit, rather than depending upon the recruitment of particular cortical regions. This observation is echoed in another recent study [17] showing similar patterns of activation across the visual system for both conscious and unconscious visual perception.
Current Biology R67
References 1. Nissen, M.J. and Bullemer, P. (1987). Attentional requirements of learning: evidence from performance measures. Cog. Psychol. 19, 1–32. 2. Willingham, D.B., Nissen, M.J. and Bullemer, P. (1989). On the development of procedural knowledge. J. Exp. Psychol. [Learn. Mem. Cogn.] 15, 1047–1060. 3. Rauch, S.L., Savage, C.R., Brown, H.D., Curran, T., Alpert, N.M., Kendrick, A., Fischman, A.J. and Kosslyn, S.M. (1995). A PET investigation of implicit and explicit sequence learning. Hum. Brain Mapp. 3, 271–286. 4. Clegg, B.A., DiGirolamo, G.J. and Keele, S.W. (1998). Sequence Learning. Trends Cog. Sci. 2, 275–281. 5. Willingham, D.B., Salidis, J. and Gabrieli, J.D.E. (2002). Direct comparison of neural systems mediating conscious and unconscious skill learning. J. Neurophysiol. 88, 1451–1460 6. Honda, M., Deiber, M.P., Ibanez, V., Pascual-Leone, A., Zhuang, P. and Hallett, M. (1998). Dynamic cortical involvement in implicit and explicit sequence learning: a PET study. Brain 121, 2159–2173 7. Grafton, S.T., Hazeltine, E. and Ivry, R. (1995). Functional anatomy of sequence learning in normal humans. J. Cog. Neuro. 7, 497–510. 8. Hazeltine, E., Grafton, S.T. and Ivry, R. (1997). Attention and stimulus characteristics determine the locus of motor-sequence encoding. A PET study. Brain 120, 123–140. 9. Grafton, S.T., Hazeltine, E. and Ivry, R (1998). Abstract and effectorspecific representations of motor sequences identified with PET. J. Neurosci. 18, 9420–9428. 10. Huettel, S.A., Mack, P.B. and McCarthy, G. (2002). Perceiving patterns in random series: dynamic processing of sequence in prefrontal cortex. Nat. Neuro. 5, 485–490. 11. Wolford, G., Miller, M.B. and Gazzaniga, M. (2000). The left hemisphere’s role in hypothesis formation. J Neurosci. 20, RC64. 12. Gomez-Beldarrain, M., Grafman, J., Pascual-Leone, A. and GarciaMonoco, J.C. (1999). Procedural learning is impaired in patient with prefrontal lesions. Neurology 52, 1853–1860. 13. Pascual-Leone, A., Wassermann, E.M., Grafman, J. and Hallett, M. (1996). The role of the dorsolateral prefrontal cortex in implicit sequence learning. Exp. Brain Res. 107, 479–485. 14. Robertson, E.M., Tormos, J.M., Maeda, F. and Pascual-Leone, A. (2001). The role of the dorsolateral prefrontal cortex during sequence learning is specific for spatial information. Cereb. Cortex 11, 628–635. 15. Verstynen, T., Le, A., Diedrichsen, J. and Ivry, R.B. (2002). Investigating the role of the dorsolateral prefrontal cortex in the acquisition and retrieval of motor sequences. Soc. Neurosci. Abstr. 67, 13. 16. Levy, R. and Goldmann-Rakic, P.S. (2000). Segregation of working memory function within the dorsolateral prefrontal cortex. Exp. Brain Res. 133, 23–32. 17. Moutoussis, K. and Zeki, S. (2002). The relationship between cortical activation and perception investigated with invisible stimuli. Proc. Natl. Acad. Sci. U.S.A. 99, 9527–9532.
