The prefrontal cortex (PFC) is critical for guiding behav- ior using working memory1. Working memory, or ���scratch- pad��� memory can be conceptualized as a form of short-term memory that is distinct from longer-term episodic (for events) or semantic (factual) memory. The information that is held within working memory is active for only a short pe- riod of time and is therefore continually updated. The PFC is thought to use the representations held ���on-line��� to guide behavior effectively in the absence of environmental cues, so freeing the organism from its dependence on the environ- ment and allowing the inhibition of inappropriate responses or distracting stimuli, thereby facilitating planning and the execution of effective organized behavior2. Animals or humans with lesions to the PFC can exhibit poor atten- tional regulation, disorganized behavior, hyperactivity and impulsivity3. These deficits can be viewed in the laboratory using tasks that challenge working memory (e.g. delayed response), planning or sequencing (e.g. Tower of London, self-ordering tasks), response inhibition (discrimination re- versal, Stroop interference task), and attention (set-shifting, gating tasks) (e.g. Refs 3���8). These PFC cognitive functions are often essential for successful organization of high-order behavior. However, in some conditions ��� acute danger, for example ��� it might be adaptive to ���shut down��� these complex, reflective operations and to allow more automatic, reflexive or habitual responses dependent on the environment to control our behavior4,5. Indeed, studies of the effects of stress on higher cognitive functioning in humans have illustrated that many of the cognitive abilities associated with the PFC are impaired by exposure to stress6���8, particularly when subjects feel that they have limited control over the stressor (e.g. Ref. 9). In contrast, highly trained or prepotent responses can actually be improved by stress6. How is it that the functions of the PFC can be altered so markedly and so rapidly? Research in animals indicates that the PFC is exquisitely sensitive to its neurochemical environment, and that catecholamines, such as dopamine (DA) and norepinephrine (NE), may have powerful effects which switch the PFC ���on-��� or ���off-line���. Although many transmitters/modulators probably con- tribute to the functional integrity of the PFC, research over the last 20 years has demonstrated the importance of the catecholamines DA and NE to the functioning of the PFC. This review describes catecholamine influences on PFC function, contrasting their beneficial and detrimental actions. The discussion focuses on research using animals and work- ing-memory tasks, which comprise the core of the work in this area. The review also focuses on actions at D1, alpha-1 and alpha-2 receptors, as infusions of drugs into PFC that act on these receptor classes have established their importance in working-memory function. A brief review of the distribu- tion of catecholamine terminals and receptors in PFC is provided in Box 1. Stress can precipitate or exacerbate many 436 Catecholamine modulation of prefrontal cortical cognitive function Amy F.T. Arnsten The prefrontal cortex (PFC) utilizes working memory to guide behavior and to release the organism from dependence on environmental cues and is commonly disrupted in neuropsychiatric disorders, normal aging, or exposure to uncontrollable stress. This review posits that the PFC is very sensitive to changes in the neuromodulatory inputs it receives from norepinephrine (NE) and dopamine (DA) systems and that this sensitivity can lead to marked changes in the working-memory functions of the PFC. While NE and DA have important beneficial influences on processing in this area, very high levels of catecholamine release, for example, during exposure to uncontrollable stress, disrupt the cognitive functions of the PFC. This fresh understanding of the neurochemical influences on PFC function has led to new treatments for cognitive disorders such as Attention Deficit Hyperactivity Disorder (ADHD), and may help to elucidate the prevalence of PFC dysfunction in other mental disorders. A.F.T. Arnsten is in the Section of Neurobiology, Yale Medical School, New Haven, CT 06520- 8001, USA. tel: +1 203 785 4431 fax: +1 203 785 5263 e-mail: amy.arnsten@ yale.edu Review A r n s t e n ��� C a t e c h o l a m i n e m o d u l a t i o n o f P F C f u n c t i o n 1364-6613/98/$ ��� see front matter �� 1998 Elsevier Science. All rights reserved. PII: S1364-6613(98)01240-6 T r e n d s i n C o g n i t i v e S c i e n c e s ��� V o l . 2 , N o . 1 1 , N o v e m b e r 1 9 9 8 TICS NOVEMBER 1998/new 19/10/98 10:50 am Page 436

neuropsychiatric disorders10, and there is increasing appre- ciation of the contribution of PFC dysfunction in neuro- psychiatric illness. Understanding the powerful effects of catecholamines on PFC function may help to elucidate and treat symptoms of PFC dysfunction. The relevance of this research to neuropsychiatric disorders is reviewed in Box 2. Catecholamines are essential to PFC working-memory function The work of Brozoski, Rosvold and Goldman11 was the first to demonstrate that catecholamines play a critical role in the modulation of the spatial working-memory functions of the PFC. Depletion of DA and NE in the PFC was produced by infusion of the catecholamine neurotoxin, 6-hydroxy- dopamine (6-OHDA), into the dorsolateral PFC of mon- keys. Infusions of 6-OHDA into the PFC destroy cat- echolamine terminals (and to a lesser extent, serotonin terminals) without effecting nonmonoaminergic cells12. Large catecholamine depletion of the PFC impaired the performance of a spatial working-memory task to the same degree as ablation of the same cortical region, illustrating the importance of catecholamine modulatory influences in this region. Animals were not impaired on a visual dis- crimination task that does not depend on the PFC, indicat- ing an effect on PFC cognitive function rather than non- specific performance deficits. The Brozoski study focused on the importance of DA to PFC function, as monkeys with large NE depletion and small DA depletion did not show deficits. However, it is now appreciated that both cat- echolamines are important to PFC function, and it is likely that both must be substantially depleted to produce marked impairment. The landmark Brozoski study has been repli- cated in rats with 6-OHDA lesions of the medial PFC (Ref. 13), and in marmosets with 6-OHDA lesions to the dorsolateral PFC (Refs 14,15). Interestingly, although cat- echolamine depletion in the dorsolateral PFC of marmosets impaired spatial working memory, it improved perform- ance of a set-shifting task14, and had no effect on a self- ordered sequencing task15, even though both tasks are sensi- tive to excitotoxic lesions of the dorsolateral PFC. These data raise the fascinating possibility that different oper- ations within the same region of the PFC may have dif- fering neurochemical requirements. This is an important area for future research. It would be particularly important to replicate these findings with infusions of catecholamine receptor antagonists into the PFC, because compensatory effects with long-term 6-OHDA lesions (e.g. receptor up-regulation) might obscure important catecholamine ac- tions for these PFC operations. Nonetheless, the present findings indicate that tasks with large working-memory demands are most likely to be disrupted by catecholamine depletion in PFC. Deficits in spatial working memory with preserved vis- ual discrimination ability have also been observed following global catecholamine depletion. For example, systemic, chronic reserpine treatment impairs delayed response per- formance without altering visual discrimination abilities in young adult monkeys16. These findings are supported by studies that have investigated the spatial working-memory performance of animals with an age-related depletion of cat- echolamines. Indeed, aged monkeys and rats with naturally occurring catecholamine depletion exhibit prominent spatial working-memory deficits17,18. Pharmacological studies in aged and in young, catecholamine-depleted animals indi- cate that both DA and NE facilitate PFC working-memory function through actions at D1 and alpha-2 adrenergic receptors, respectively (Fig. 1A). DA enhances PFC working-memory function via D1 receptor stimulation Brozoski et al. illustrated the importance of DA mecha- nisms to PFC working-memory function by showing that the mixed D1/D2 agonist, apomorphine, could ameliorate delayed alternation deficits in their catecholamine-depleted monkeys11. This interpretation was reinforced by electro- physiological findings in monkeys performing a working- memory task, as iontophoretic administration of DA onto PFC neurons enhanced memory-related cell firing during the critical delay period19. DA can also excite PFC neurons that respond to the cue signal and/or to the go signal in- forming the monkey it is time to respond for reward20. These data are quite consistent with recordings from the cell bodies of dopaminergic neurons in monkeys performing a working-memory task, where the DA cells fire in response to cues associated with reward21,22. Moreover, in highly trained monkeys, the DA neurons fire to the visuo-spatial cue signal. One thus could speculate that DA release is en- hanced in the PFC at the initiation of each trial by the cue signal, in time to modulate cue-related, delay-related and go-related activity. Note that this speculation assumes long-acting effects of DA on PFC cells, that is, over many seconds, presumably via second-messenger actions. Research in monkeys and rodents has now established the importance of the D1 receptor ���family��� in the regulation of PFC function. Recordings from PFC neurons in mon- keys performing a working-memory task first suggested that D1 receptors may be more influential than D2 receptors for working-memory function, as mixed D1/D2 family an- tagonists suppressed delay-related activity of PFC neurons, while a more selective antagonist acting at D2 receptors was ineffective19. This hypothesis was supported by behavioral studies that showed that infusions of selective D1/D5, but not D2 /D3 antagonists into the PFC markedly impaired spatial working memory, as measured by the oculomotor delayed-response task23. The impairment seen with D1/D5 receptor-antagonist infusion has now been replicated in rat PFC (Ref. 24), and has been found with systemic D1/D5 receptor-antagonist administration in monkeys25 and rats26 (Fig. 2). Low doses of a D2/D3/D4 agonist that inhibit DA release similarly impair performance on the delayed- response task27. Conversely, working memory can be im- proved by low dose D1/D5 agonist treatment in aged mon- keys with naturally occurring DA depletion25,28. Selective, full D1/D5 agonists have only recently become available for research in animals28, and are not yet available for human usage thus, less is known about D1/D5 influences on cog- nitive function in humans. However, the mixed D1/D2 family agonist, pergolide, appears to produce more con- sistent improvement in spatial working memory in hu- mans than the agonist from the D2 family, bromocriptine, A r n s t e n ��� C a t e c h o l a m i n e m o d u l a t i o n o f P F C f u n c t i o n 437 T r e n d s i n C o g n i t i v e S c i e n c e s ��� V o l . 2 , N o . 1 1 , N o v e m b e r 1 9 9 8 Review TICS NOVEMBER 1998/new 19/10/98 10:50 am Page 437