Motivated Memory: Integrating Cognitive and Affective Neuroscience

Abstract

A growing body of literature indicates that motivation can critically shape long-term memory formation in the service of adaptive behavior. In the present chapter, we review recent cognitive neuroscience evidence of motivational influences on memory, with a focus on anatomical pathways by which neuromodulatory networks support encoding-related activity in distinct subregions of the medial temporal lobe. We argue that engagement of distinct neural circuits as a function of motivational context at encoding leads to formation of different memory representations, supporting different patterns of adaptive behavior. We present a novel neurocognitive model, the Interrogative/Imperative model of information-seeking, to account for pursuit of learning goals. Interrogative or imperative modes of information-seeking are often, but not necessarily, associated with approach or avoidance motivation, respectively. We also discuss additional influences on motivated memory encoding, including intrinsic motivation, curiosity, choice and cognitive control processes. Taken together, this body of research suggests that the nature of memory representations depends on an individual’s neurophysiological response to, rather than extrinsic qualities of, a given motivational manipulation or context at the time of encoding. Finally, we discuss potential applications of these research findings to real-life educational settings and directions for future research. INTEGRATING COGNITIVE AND AFFECTIVE NEUROSCIENCE 3 Motivated Memory: Integrating Cognitive and Affective Neuroscience Motivation is critical to learning and memory, and there is widespread use of strategies to motivate learning in the classroom, many of which rely on intuition. However, it is only recently that empirical neuroscience research has begun to examine the large repertoire of motivated behaviors and memory processes central to education, positioning itself to provide evidencebased solutions. Early work in neuroscience focused on relatively simple forms of associative learning. These included Pavlovian stimulus-stimulus learning (where a motivationally significant stimulus is associated with a previously neutral stimulus; for example, the sound of a bell with a food reward) and instrumental learning stimulus-response learning (where the strength of a behavior is modified by rewarding or punishing consequences; for example, learning to press a button to receive a food reward). Using such models has yielded a rich psychology and neuroscience literature on motivation in associative learning. More recently, these investigations have been augmented by new lines of cognitive neuroscience research that address a common intuition among educators: that motivation plays a critical role in how information is learned and encoded into long-term memory. The recognition that motivational influences can influence long-term memory processing by the medial temporal lobe represents an important advance in memory research. Learning and memory over extended timescales (i.e., multiple days or years) is essential to adaptive behavior and cannot be accounted for by the Pavlovian or instrumental associative learning mechanisms historically studied in the context of motivation. In the present chapter, we review emerging cognitive neuroscience evidence regarding motivational effects on learning and memory formation. We discuss several factors important to motivational contexts and their impact on INTEGRATING COGNITIVE AND AFFECTIVE NEUROSCIENCE 4 neural activity and cognition, including incentive salience, expectation, extrinsic vs. intrinsic motivation, curiosity, and choice. Although this new research area opens many questions, based on present evidence we argue that distinct motivational states serve as adaptive contexts for learning, engaging distinct neural circuitry to support memory encoding, and thus leading to distinct forms of memory representation. Importantly, present evidence suggests that it is an individual’s neurophysiological response to a motivational manipulation that is critical in determining the nature of the encoded memory representation, rather than extrinsic qualities of the incentive. Finally, we discuss the implications of these basic scientific findings regarding motivation and learning in applied settings, such as educational or managerial environments. Motivation and Memory Encoding: Core Neural and Psychological Substrates The medial temporal lobes (MTL) of the brain have long been recognized as critical to encoding and retrieval of declarative long-term memory (Squire, Zola-Morgan, & Stuart, 1991; Tulving & Markowitsch, 1998). The MTL is comprised of sub-structures that play complementary, yet distinct roles in these processes; these structures include the hippocampus and surrounding cortical regions: perirhinal, parahippocampal, and entorhinal cortices (reviewed in Davachi, 2006). Decades of research in both animal and human models suggest that the hippocampus plays an essential role in binding elements of an episode into an interrelated, multimodal long-term memory (Eichenbaum, 2000; Tulving, 1985, 2002). More recent evidence has refined this account, suggesting that overlying cortical regions represent distinct aspects or features of an episode to be bound together: perirhinal cortex selectively supports memory for items previously encountered, and parahippocampal cortex supports memory for the environmental context where the items were encountered. The hippocampus thus binds item and INTEGRATING COGNITIVE AND AFFECTIVE NEUROSCIENCE 5 context memory together to produce a coherent memory episode (Davachi, 2006; Konkel & Cohen, 2009; Ranganath, 2010). Largely distinct from the research literature investigating MTL and long-term memory, a separate body of affective neuroscientific research has sought to characterize the modulatory neurotransmitter dopamine and its effects on motivated behavior. Widespread evidence suggests that the mesolimbic and mesocortical dopamine pathways, which primarily originate in the ventral tegmental area (VTA) and associated nuclei in the midbrain and project widely to regions in limbic and cerebral cortex, critically support motivated pursuit of a broad range of rewards, such as food and sex (Olds & Milner, 1954; Willner & Scheel-Krüger, 1991). Given the opportunity to directly self-stimulate VTA (i.e., through a button press), rodents forego other biologically-relevant rewards to continue self-stimulation, providing direct evidence for a critical role of the VTA in motivation and reinforcement (Olds & Milner, 1954). Neuroimaging studies in humans using functional magnetic resonance imaging (fMRI), have likewise observed increased mesolimbic activity during anticipation and pursuit of reward relative to non-reward outcomes, typically using secondary rewards such as money (Carter, MacInnes, Huettel, & Adcock, 2009; Knutson, Taylor, Kaufman, Peterson, & Glover, 2005). Functional relationships between the mesolimbic dopamine system and the MTL are well positioned to support motivated learning. Animal studies have long demonstrated anatomical connectivity between the VTA and hippocampus (Amaral & Cowan, 1980; Samson, Wu, Friedman, & Davis, 1990). Novel stimuli can elicit midbrain dopamine neuron activity, and dopamine released in the hippocampus stabilizes long-term potentiation (LTP) potentially supporting learning of new information (reviewed in Lisman & Grace, 2005), including single1 Long-term potential (LTP) refers to a persistent increase in synaptic strength based on recent patterns of neuronal firing at that synapse. INTEGRATING COGNITIVE AND AFFECTIVE NEUROSCIENCE 6 trial learning (Neugebauer, Korz, & Frey, 2009; O’Carroll, Martin, Sandin, Frenguelli, & Morris, 2006; reviewed further in Shohamy and Adcock, 2010). Despite these findings in learning and memory and the robust link between dopamine and motivation, the potential effects of motivational manipulations on encoding into long-term memory and their supporting neural circuitry have only recently begun to be investigated. Wittmann et al., (2005) examined incidental memory (with a 3-week delay between encoding and retrieval) for reward predictive and non-reward predictive picture stimuli, and observed a subsequent memory benefit for stimuli that predicted reward over those that did not. The researchers found that this memory benefit was associated with enhanced activity both in dopaminergic midbrain and in the hippocampus at the time of encoding. Likewise, a seminal experiment from our laboratory (Adcock, Thangavel, Whitfield-Gabrieli, Knutson, & Gabrieli, 2006) examined neural activity during memory encoding as a function of a reward incentive manipulation. In contrast to incidental memory encoding in the study by Wittmann and colleagues (2005), participants intentionally encoded picture stimuli in anticipation of receiving monetary incentive (signaled prior to each picture stimulus with a high vs. low value reward cue) for successfully remembering each in a recognition test 24 hours later. Recognition memory was superior for stimuli associated with highvs. low-value cues. Additionally, encoding of high (vs. low) value stimuli was associated with enhanced anticipatory activity in the VTA as well as the hippocampus. Importantly, on a trial-by-trial basis, functional connectivity between the VTA and hippocampus at the time of encoding predicted subsequent memory success. In addition to mesolimbic and MTL regions, research has indicated a role for higher cortical regions in motivated memory encoding. The lateral prefrontal cortex (PFC) is robustly innervated by dopamine (Goldman-Rakic & Friedman, 1991; Sawaguchi & Goldman-Rakic, INTEGRATING COGNITIVE AND AFFECTIVE NEUROSCIENCE 7 1991), may provide informational input to dopaminergic midbrain to support adaptive behavior (Ballard et al., 2011), and plays a fundamental role in supporting cognitive control processes (Miller & Cohen,