Dynamic neuroplasticity and the automation of motivated behavior.

Extraordinary behavioral flexibility is necessary in order to meet energy needs. An animal must know when and where to search for food, guard itself and its offspring against predators, and continuously adjust foraging behavior as the changing environment dictates. These pressures have selected for the ability to accomplish numerous functions simultaneously. This is particularly necessary during the execution of motivated behaviors (behaviors that have initiation, procurement, and consummatory phases) (Swanson 2000), and is facilitated by habitual performance. Specifically, while motivated learning initially requires significant cognitive effort, with practice, complex tasks can be performed quite automatically. In a recent study, Hernandez et al. (2006) examined the neurobiology of this transition, from early learning to automatic responding, based on the hypothesis that the transition is mediated by plasticity within neural circuitry regulating motivated learning. Hernandez et al. assessed neural plasticity anatomically using in situ hybridization to measure changes in the expression of two immediate early genes important for learning and memory: Zif268 and Homer 1a. Zif268 is a transcription factor involved in cellular and behavioral assays of learning (Knapska and Kaczmarek 2004). Homer 1a is a component of the postsynaptic density of glutamate synapses (de Bartolomeis and Iasevoli 2003). This is especially important given the role played by glutamate transmission in learning (Gubellini et al. 2004; Kelley 2004). In addition to a general involvement in learning and memory, Zif268 and Homer 1a are also specifically important for motivated learning (Thomas et al. 2003; Lominac et al. 2005; Petrovich et al. 2005; Szumlinski et al. 2006). These genes were therefore chosen to examine whether plasticity within motivational brain circuitry is associated with the transition from initial motivated responding to automatic responding. Rats learned an operant task where lever presses produced sugar pellets. In one group, gene expression was measured after several days of training, in which behavior was reliable but responding had not reached its maximum. These subjects were referred to as “early learners.” In a separate group, gene expression was measured several weeks later, well after the behavioral measure had reached its maximum level, and these subjects were referred to as “overtrained.” Control subjects were exposed to the same conditions; however, lever pressing had no programmed consequences. Relative to controls, early learners showed significantly higher expression levels of Zif268 and Homer 1a in several frontal cortical regions (including anterior cingulate, orbitofrontal, and prefrontal). In contrast with the early learners, overtrained subjects had significantly lower Homer 1a expression in these cortical regions compared with controls (there was no difference for Zif268). This reversal was only found in the frontal cortex and not in other areas where gene expression was elevated in early learners, such as the dorsomedial striatum, hippocampus, or sensory and motor cortex. These findings suggest decreased cortical influence during automatic responding. Hernandez et al. (2006) then directly compared levels of gene expression in overtrained subjects to early learners. Overtrained rats had significantly higher Homer 1a expression in the ventrolateral striatum compared to early learners. Further, there was a positive correlation between the number of lever presses and Homer 1a expression in the ventrolateral striatum. These findings suggest that this specific portion of the striatum plays an important role in automation of behavior. And together, data presented in Hernandez et al. suggest a “dynamic shift” in the expression of plasticity-related immediate early gene expression, in which there is decreased cortical and increased striatal activity as a specific motivated output becomes very well established. Importantly, this “cortex to striatum” shift may be unique to motivational learning, as a fundamentally distinct mechanism has been identified for spatial and contextual memory storage. Several recent studies have used similar neuroanatomical methods, as used by Hernandez et al. (2006) to convincingly demonstrate that recent spatial memories involve processing within the hippocampus, whereas remote spatial memories (∼30 d after learning) require cortical processing (Frankland et al. 2004; Maviel et al. 2004; Frankland and Bontempi 2005). These studies demonstrate differential neural regulation of learning and memory depending on the behavior being examined. Dynamic neuroplasticity mediating transitions from early learning to automatic responding is important for understanding the evolution of motivated behavior. Approach and avoidance behaviors (Adler 1966; Zhang et al. 2005), as well as operant learning (Brembs 2003; Hawkins et al. 2006), are ancient processes predating neocortex. For food seeking in mammals, the ability to distinguish between positive and negative valence is processed in phylogenetically older brain regions in the midand hindbrain (Grill and Norgren 1978), areas expressing very low levels of plasticity-related genes (Kelley 2004). However, these regions are highly interconnected to the more phylogenetically recent structures, such as neocortex and subcortical forebrain structures like the striatum. These regions possess the neuroplastic capabilities to efficiently meet energy needs by orchestrating the appropriate influence over hard-wired feeding areas, depending on level of experience of a particular motivated output. Specifically, with practice a complex behavior can be executed with only a minimal burden of awareness, because it is under striatal regulation. This leaves the cortex free to process additional features of the foraging experience, such as sensing danger or changes in the environment that may signal more promising foraging opportunities. If true, such shifts in corticostriatal regulation of motivated behavior can explain the adaptive benefit of habitual behavior. Perhaps the most significant aspect of highly plastic capabilities of the coritcostriatal system mediating motivated learning is that this system is extremely sensitive to drugs of abuse (Nestler 2004). Drug addiction is widely conceptualized as a disorder of learning and memory (Di Chiara 1999; Hyman 2005), and increasing evidence suggests that compulsive drug seeking is 1Corresponding author. E-mail [email protected]; fax (919) 962-2537. Article is online at http://www.learnmem.org/cgi/doi/10.1101/lm.398806. Commentary