- Email: [email protected]
All the King's Horses and All the King's Men: Putting the Brain Back Together Again
Brain and Cognition 42, 7–9 (2000) doi:10.1006/brcg.1999.1146, available online at http://www.idealibrary.com on
All the King’s Horses and All the King’s Men: Putting the Brain Back Together Again Joseph B. Hellige University of Southern California
‘‘The time has come to put the brain back together again. In fact, one of the most important challenges facing cognitive neuroscience is to account for the emergence of unified processing from a brain consisting of a variety of processing subsystems’’ (Hellige, 1993, p. 206). I made the above observation a few years ago in a chapter on the interaction of the two cerebral hemispheres. I noted that decades of research dealing with differences between the left and right cerebral hemispheres had resulted in the brain being taken apart in a metaphoric sense, with analysis and conceptualization emphasizing the two hemispheres as separate processing systems, each with its own abilities and propensities. There is no doubt that cerebral laterality is a fundamental property of brain and behavior in humans and other species. However, in order to understand the unity of perception, cognition, emotion, intelligence, and action we must also understand the variety of ways in which the two hemispheres interact and the biological mechanisms that support those interactions. Indeed, studies in the intervening years have shed light on the roles of the corpus callosum and subcortical structures, the various forms of interhemispheric interaction from an information processing perspective and the implications for theories of attention (for summary and discussion see Banich, 1998). But, we are still far from a satisfactory understanding of the brain’s unified processing strategy. The metaphoric breaking apart of the brain is even more characteristic of the fine-grained modularity that characterizes much of contemporary cognitive neuroscience. In the attempt to ‘‘carve nature at its joints,’’ experimental techniques have been developed to dissect behavioral tasks into successively smaller and more basic processing components, with the goal of discovering components that have a good deal of generality across tasks. At the same Address correspondence and reprint requests to Joseph B. Hellige, Department of Psychology, University of Southern California, Los Angeles, CA 90089-1061. E-mail: hellige@ usc.edu. 7 0278-2626/00 $35.00 Copyright 2000 by Academic Press All rights of reproduction in any form reserved.
8
MILLENNIUM ISSUE
time, advances in neuroimaging and other technologies have allowed some of these processes to be associated with more and more precisely identified and circumscribed brain areas. The identification of the resulting mind/brain modules is a significant achievement and essential to understanding brain and cognition, with much good work remaining to be done along these lines. But, as the modular structure of mind and brain becomes even clearer, an issue of increasing importance will be understanding the ways in which modules interact, at descriptive, computational, and biological levels. This will be especially true for understanding aspects of cognition that are of particular interest to the lay public, things such as intelligence, wisdom, and consciousness. To be sure, the issue of interaction among processing subsystems or modules is not new. For example, at the level of entire hemispheric systems, I noted earlier the importance of interhemispheric interaction. For an example at a more fine-grained level, it is useful to consider the so-called ‘‘binding problem’’ in visual perception. Various properties of an object such as shape, color, location, and so forth are processed by different subsystems associated with different areas of the cortex. This being the case, when there is more than one object present, how do the various properties associated with a single object become bound to each other? And, at an even finer-grained level, within the shape-processing subsystem, how are various features bound to each other in the appropriate spatial configuration? Although these sorts of binding problems are far from being solved, various computational models have successfully represented conjunctions of attributes by synchronizing the firing rates of units representing those attributes. While it remains to be determined whether temporal synchrony is the mechanism actually used by the brain in each case, such a mechanism is at least neurologically plausible. Understanding how processing subsystems interact also requires us to acknowledge that such subsystems can be specified at many levels of scale. So far, I have emphasized the importance of learning how the subsystems within a particular level of scale interact with each other. But, that is not enough. Complete understanding of brain and cognition also demands a web of causality that shows how the different levels of scale interact with each other. For example, we must address such questions as whether and how we can build a coherent theory of laterality from knowledge of modularity at a more fine-grained level, considering whether there are any emergent properties that can only be understood by interrogating the system at the more coarse level. Likewise, we want to be able to specify processing at the modular level by understanding the neuronal networks that make them up. Dealing with these issues calls for a new kind of integrated approach to research and theory and a breadth of perspective that is unusual in contemporary work and graduate training. One illustrative first approximation is provided by Stephen Kosslyn’s theory of visual cognition and imagery (e.g., Kosslyn, 1996). While many specific details of the theory may be incorrect, the approach taken and the principles that are articulated for moving among levels
MILLENNIUM ISSUE
9
of analysis provide a framework for the kind of holistic thinking that the study of brain and cognition will require in the years to come. The fact that we can describe the elements of mind and brain at different levels of scale also leads to considerations regarding the level of scale that is most relevant for answering different sorts of questions for different audiences. Part and parcel of the scientific approach is the dissection of mind and brain into successively smaller, more basic, and more general units. This does not necessarily mean, however, that the most fine-grained or reductionistic level of scale will provide the level of explanation that is most useful for all purposes or all phenomena. For example, although we may eventually understand the processes involved in reading at the neuronal or even molecular level, for many purposes it may be preferable to emphasize an account at the level of larger mind/brain units. Consequently, as we become more successful at understanding and integrating the various levels from behavior to molecules, we must articulate the advantages and limitations of explanations at each level. These issues that arise in putting the brain back together are interwoven with a number of others that will become increasingly more important. For example, I doubt that we can really understand how the modules and levels of mind and brain interact with each other in contemporary humans without understanding the context in which mind and brain emerged. This requires comparisons across species as well as consideration of development across both evolutionary time and across the life span of an individual from womb to tomb (see Hellige, 1993, and Kosslyn, 1996, for discussion related to hemispheric asymmetry and to visual cognition, respectively). It also requires a deeper understanding of the interplay among such varied things as genes, the environment, social interaction, and culture. In seeking this deeper, holistic understanding, we should be mindful that, as the biologist Edward Wilson reminds us, putting things back together is as much a part of our enterprise as breaking them apart. ‘‘While it is true that science advances by reducing phenomena to their working elements—by dissecting brains into neurons, for example, and neurons into molecules—it does not aim to diminish the integrity of the whole. On the contrary, synthesis of the elements to recreate their original assembly is the other half of scientific procedure. In fact, it is the ultimate goal of science’’ (Wilson, 1998, p. 230). REFERENCES Banich, M. T. 1998. Integration of information between the cerebral hemispheres. Current Directions in Psychological Science, 7, 32–37. Hellige, J. B. 1993. Hemispheric asymmetry: What’s right and what’s left. Cambridge, MA: Harvard University Press. Kosslyn, S. M. 1996. Image and brain: The resolution of the imagery debate. Cambridge, MA: MIT Press. Wilson, E. O. 1998. Consilience: The unity of knowledge. New York: Knopf.
All the King’s Horses and All the King’s Men: Putting the Brain Back Together Again Joseph B. Hellige University of Southern California
‘‘The time has come to put the brain back together again. In fact, one of the most important challenges facing cognitive neuroscience is to account for the emergence of unified processing from a brain consisting of a variety of processing subsystems’’ (Hellige, 1993, p. 206). I made the above observation a few years ago in a chapter on the interaction of the two cerebral hemispheres. I noted that decades of research dealing with differences between the left and right cerebral hemispheres had resulted in the brain being taken apart in a metaphoric sense, with analysis and conceptualization emphasizing the two hemispheres as separate processing systems, each with its own abilities and propensities. There is no doubt that cerebral laterality is a fundamental property of brain and behavior in humans and other species. However, in order to understand the unity of perception, cognition, emotion, intelligence, and action we must also understand the variety of ways in which the two hemispheres interact and the biological mechanisms that support those interactions. Indeed, studies in the intervening years have shed light on the roles of the corpus callosum and subcortical structures, the various forms of interhemispheric interaction from an information processing perspective and the implications for theories of attention (for summary and discussion see Banich, 1998). But, we are still far from a satisfactory understanding of the brain’s unified processing strategy. The metaphoric breaking apart of the brain is even more characteristic of the fine-grained modularity that characterizes much of contemporary cognitive neuroscience. In the attempt to ‘‘carve nature at its joints,’’ experimental techniques have been developed to dissect behavioral tasks into successively smaller and more basic processing components, with the goal of discovering components that have a good deal of generality across tasks. At the same Address correspondence and reprint requests to Joseph B. Hellige, Department of Psychology, University of Southern California, Los Angeles, CA 90089-1061. E-mail: hellige@ usc.edu. 7 0278-2626/00 $35.00 Copyright 2000 by Academic Press All rights of reproduction in any form reserved.
8
MILLENNIUM ISSUE
time, advances in neuroimaging and other technologies have allowed some of these processes to be associated with more and more precisely identified and circumscribed brain areas. The identification of the resulting mind/brain modules is a significant achievement and essential to understanding brain and cognition, with much good work remaining to be done along these lines. But, as the modular structure of mind and brain becomes even clearer, an issue of increasing importance will be understanding the ways in which modules interact, at descriptive, computational, and biological levels. This will be especially true for understanding aspects of cognition that are of particular interest to the lay public, things such as intelligence, wisdom, and consciousness. To be sure, the issue of interaction among processing subsystems or modules is not new. For example, at the level of entire hemispheric systems, I noted earlier the importance of interhemispheric interaction. For an example at a more fine-grained level, it is useful to consider the so-called ‘‘binding problem’’ in visual perception. Various properties of an object such as shape, color, location, and so forth are processed by different subsystems associated with different areas of the cortex. This being the case, when there is more than one object present, how do the various properties associated with a single object become bound to each other? And, at an even finer-grained level, within the shape-processing subsystem, how are various features bound to each other in the appropriate spatial configuration? Although these sorts of binding problems are far from being solved, various computational models have successfully represented conjunctions of attributes by synchronizing the firing rates of units representing those attributes. While it remains to be determined whether temporal synchrony is the mechanism actually used by the brain in each case, such a mechanism is at least neurologically plausible. Understanding how processing subsystems interact also requires us to acknowledge that such subsystems can be specified at many levels of scale. So far, I have emphasized the importance of learning how the subsystems within a particular level of scale interact with each other. But, that is not enough. Complete understanding of brain and cognition also demands a web of causality that shows how the different levels of scale interact with each other. For example, we must address such questions as whether and how we can build a coherent theory of laterality from knowledge of modularity at a more fine-grained level, considering whether there are any emergent properties that can only be understood by interrogating the system at the more coarse level. Likewise, we want to be able to specify processing at the modular level by understanding the neuronal networks that make them up. Dealing with these issues calls for a new kind of integrated approach to research and theory and a breadth of perspective that is unusual in contemporary work and graduate training. One illustrative first approximation is provided by Stephen Kosslyn’s theory of visual cognition and imagery (e.g., Kosslyn, 1996). While many specific details of the theory may be incorrect, the approach taken and the principles that are articulated for moving among levels
MILLENNIUM ISSUE
9
of analysis provide a framework for the kind of holistic thinking that the study of brain and cognition will require in the years to come. The fact that we can describe the elements of mind and brain at different levels of scale also leads to considerations regarding the level of scale that is most relevant for answering different sorts of questions for different audiences. Part and parcel of the scientific approach is the dissection of mind and brain into successively smaller, more basic, and more general units. This does not necessarily mean, however, that the most fine-grained or reductionistic level of scale will provide the level of explanation that is most useful for all purposes or all phenomena. For example, although we may eventually understand the processes involved in reading at the neuronal or even molecular level, for many purposes it may be preferable to emphasize an account at the level of larger mind/brain units. Consequently, as we become more successful at understanding and integrating the various levels from behavior to molecules, we must articulate the advantages and limitations of explanations at each level. These issues that arise in putting the brain back together are interwoven with a number of others that will become increasingly more important. For example, I doubt that we can really understand how the modules and levels of mind and brain interact with each other in contemporary humans without understanding the context in which mind and brain emerged. This requires comparisons across species as well as consideration of development across both evolutionary time and across the life span of an individual from womb to tomb (see Hellige, 1993, and Kosslyn, 1996, for discussion related to hemispheric asymmetry and to visual cognition, respectively). It also requires a deeper understanding of the interplay among such varied things as genes, the environment, social interaction, and culture. In seeking this deeper, holistic understanding, we should be mindful that, as the biologist Edward Wilson reminds us, putting things back together is as much a part of our enterprise as breaking them apart. ‘‘While it is true that science advances by reducing phenomena to their working elements—by dissecting brains into neurons, for example, and neurons into molecules—it does not aim to diminish the integrity of the whole. On the contrary, synthesis of the elements to recreate their original assembly is the other half of scientific procedure. In fact, it is the ultimate goal of science’’ (Wilson, 1998, p. 230). REFERENCES Banich, M. T. 1998. Integration of information between the cerebral hemispheres. Current Directions in Psychological Science, 7, 32–37. Hellige, J. B. 1993. Hemispheric asymmetry: What’s right and what’s left. Cambridge, MA: Harvard University Press. Kosslyn, S. M. 1996. Image and brain: The resolution of the imagery debate. Cambridge, MA: MIT Press. Wilson, E. O. 1998. Consilience: The unity of knowledge. New York: Knopf.