Welcome to

Mind Matters

where top researchers in neuroscience, psychology, and psychiatry explain and discuss the findings and theories driving their fields. Join us via the "comments" link at the end of the post. -- David Dobbs, Editor, Mind Matters
This week:

Modules, Networks, and the Brain's Need to Heed



by David Dobbs
Editor, Mind Matters
Two perennial polarities beloved by brain geeks -- networks versus modules and top-down versus bottom-up attention -- get linked in this week's essay, in which UC Berkeley's Mark D'Esposito reviews an imaging study of how monkeys use their brains to direct their attention. The results, suggests D'Esposito, add threads to vital strands of neuroscientific thought.

How the Brain Coordinates its Efforts to Pay Heed

Mark D'Esposito
Helen Wills Neuroscience Center
University of California, Berkeley

How does the brain organize its work? And how does it heed what it needs to heed? Theories of brain organization focus on two distinct but complementary principles of brain organization: modularity, the existence of brain regions with specialized functions, and network connectivity, the integration of information from various brain regions that results in organized behavior. In the study under review here, the modular and network models appear to play specialized roles in directing the attention of monkeys seeking certain visual targets through either "top-down" or "bottom-up" attentional strategies. Modules versus networks In the modules-versus-network debate, modularity is probably the simpler brain model to understand. Clinical observation of individuals with brain damage, as well as brain-imaging studies (functional MRIs, or fMRIs) of healthy individuals, demonstrate that certain brain regions control specific cognitive processes, such as the ability to produce speech. For instance, in patients with nonfluent aphasia, which creates a selective inability to speak, comprehension of spoken language remains intact. In 1861 Paul Broca observed that damage to the left frontal lobe in an autopsied brain had produced nonfluent aphasia. Modern brain-imaging studies of patients with strokes to this area (now known as "Broca's area") confirmed Broca's theory. Moreover, fMRIs of healthy individuals reveal that the left frontal lobe is activated when subjects generate speech. Of course, that some brain areas specialize in certain functions does not exclude the possibility that those areas are also part of larger networks of brain regions communicating with one another. Although the modular model may accurately describe many cognitive functions, it is insufficient to explain complex cognitive processes that cannot be localized to isolated brain regions. It is unlikely that our ability to get the gist of a conversation, for instance, is the work of a single specialized brain module. Such complex behavior more likely arises from interactions between brain regions through network connectivity. In his 1995 book Memory in the Cerebral Cortex: An Empirical Approach to Neural Networks in the Human and Nonhuman Primate, UCLA neurologist Joaquin Fuster began an argument he extended in his 2002 work Cortex and Mind: Unifying Cognition. According to Fuster, new studies of brain networks have led to a "revolution in contemporary neuroscience." He contends that the empirical shift from a reductionist modular model to a holistic network model offers promise of accomplishing our long-term goal of resolving the mind-brain question. Fuster's conception of a network model of brain function includes several key notions: (1) Cognitive information is represented in wide, overlapping and interactive brain networks. (2) Such networks develop on a core of organized modules of elementary sensory and motor functions, to which they remain connected. (3) The cognitive code is a relational code, based on connectivity between discrete brain regions. (4) The code's diversity and specificity derive from the myriad possible combinations of those brain regions. (5) Any brain region can be part of many networks and thus of many percepts, memories, items of experience or personal knowledge. (6) A given brain network can serve several cognitive functions. (7) Cognitive functions consist of functional interactions within and between brain networks. A division of labor ... In "Top-Down Versus Bottom-Up Control of Attention in the Prefrontal and Posterior Parietal Cortices" (click here for pdf download), published in Science this year, Timothy Buschman and Earl Miller, both of MIT, add to this model by exploring the neural mechanisms underlying both functional specialization and functional integration. Buschman and Miller investigated how the brain allows us to give volitional attention to something in our environment, such as looking for our keys (top-down attention), versus automatically attending to something salient (or attention-grabbing), such as a fire alarm (bottom-up attention). To study these processes, the researchers focused on two brain regions known to be involved in attentional processes, the frontal and parietal lobes. Using monkeys, Buschman and Miller recorded from neurons in these brain regions while the monkeys located a visual target on a computer screen. The target was always randomly located in an array of four stimuli, but there were two sets of conditions: In the first, the "pop-out" task, the target object would be clearly different from the nontarget stimuli (for example, it was not only a different color but also a different orientation), making the target more conspicuous. In the second test, the "search" task, the target stimulus would match some of the nontarget stimuli in at least a few dimensions (for instance, it might have the same color but not the same orientation). Because the target stimulus in this latter test was not salient, the monkeys had to rely on their memory of the desired target's appearance as they looked for it. As expected, it took longer for the monkeys to find this second type of target. Buschman and Miller found that frontal-lobe neurons were first to find the target during the search task, whereas parietal lobe neurons were first to find the target during the pop-out task. In other words, selection of a more obscure target (top-down attention) may be mediated by the frontal lobes, whereas fast selection of a highly salient target (bottom-up attention) may be mediated by the parietal lobes. These findings suggest that the frontal and parietal lobes may mediate different cognitive processes consistent with functional specialization. ... and cooperative ventures The researchers also investigated how these two brain regions communicated with each other during the two attention tasks by measuring the degree of synchrony between neuronal activity in each region. (Such synchrony -- a rough alignment of electrical wave patterns emitted from different brain areas -- is thought to facilitate or indicate communication and cooperation among brain regions. See Gyorgy Buzsaki's Rhythms of the Brain.) Synchrony between frontal and parietal regions was stronger in lower frequencies during the search task and in higher frequencies during the pop-out task, suggesting that synchronous activity between brain regions may increase the effectiveness of communication between these regions. Also, different modes of attention (for example, top-down as opposed to bottom-up) may emphasize synchrony at different frequencies. Buschman and Miller proposed that the increase in low-frequency synchrony during the search task could reflect a "broadcast" of top-down signals across the entire brain, whereas higher-frequency synchrony may support the local interactions between brain regions. A methodological advance These results add detail to the picture neuroscience is getting of how brain networks operate. And the study's methodological advance -- the ability to record simultaneously from two distant brain regions in an awake and active monkey -- offers a significant new tool for studying brain network connectivity. (In humans, functional magnetic resonance imaging, or fMRI, can also explore interactions between brain regions, as it simultaneously records correlates of neural activity throughout the entire functioning brain.) Over the next decade, hundreds of roughly similar monkey neurophysiology and human imaging studies will be performed in neuroscience laboratories around the world. From these studies, the pieces of the puzzle will continue to fall into place, allowing us to make significant strides toward answering the ultimate question: How does the brain work? Mark D'Esposito is Professor of Neuroscience and Psychology at the University of California, Berkeley, where he the Director of the Henry H. Wheeler Jr. Brain Imaging Center at the Helen Wills Neuroscience Institute. He studies working memory, frontal lobe function, and cognitive neuroscience. He also maintains a practice in clinical neurology. -- Edited by David Dobbs at 11/30/2007 11:26 AM -- Edited by David Dobbs at 11/30/2007 11:28 AM