- Email: [email protected]
Ruthless reductionism and social cognition
Available online at www.sciencedirect.com
Journal of Physiology - Paris 101 (2007) 230–235 www.elsevier.com/locate/jphysparis
Ruthless reductionism and social cognition John Bickle Department of Philosophy and Neuroscience Graduate Program, University of Cincinnati, Cincinnati, OH, USA
Abstract Social cognition appears to present phenomena that ‘‘ruthlessly reductive” molecular and cellular neuroscience cannot fruitfully investigate or explain. This is because the causes of such phenomena are distal and external not only to the molecular machinery of individual neurons, but to individual brains. However, the ‘‘reductionist’s epiphany” insists that to the extent that we understand the specific molecular mechanisms that underlie phenomena upon which most or all social cognition depends, we can be sure that molecular mechanisms for the broader phenomena can be found using standard experimental methods from molecular and cellular cognition. Furthermore, social recognition memory consolidation is required for virtually all types of social cognition, and its specific molecular mechanisms have now been uncovered experimentally. These same molecular mechanisms obtain across a wide variety of divergent species (from invertebrates to vertebrates). Thus we can expect to find the molecular mechanisms of the broader social cognitive functions that must ‘‘plug into” these specific molecular mechanisms, despite these functions’ typically distal, external initial causes. This conclusion rests on explicit scientific facts, not just on some vague philosophical commitment to physicalism about mind. Ó 2008 Elsevier Ltd. All rights reserved. Keywords: Reductionist’s epiphany; Social recognition memory consolidation; cAMP-PKA-CREB pathway; CREBad mutant mice; Molecular evolution
1. Introduction: Ruthless reductionism in recent neuroscience ‘‘Ruthless reductionism” informs key branches of contemporary neuroscience. Its features are nicely illustrated in a passage from the introductory chapter of Kandel, Schwartz, and Jessell’s state-of-the-art textbook, Principles of Neural Science: This book . . . describes how neural science is attempting to link molecules to mind—how proteins responsible for the activities of individual nerve cells are related to the complexity of neural processes. Today it is possible to link the molecular dynamics of individual nerve cells to representations of perceptual and motor acts in the brain and to relate these internal mechanisms to observable behavior. (2001, 3–4) This passage nicely illustrates that the explanatory target of these investigations is mind, not some ersatz laboratory stand-in. The explanations and experimental manipulations
E-mail address: [email protected] 0928-4257/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.jphysparis.2007.11.009
have now reached into molecular mechanisms internal to individual neurons. That this passage occurs in one of neuroscience’s principal current textbooks indicates the centrality of ruthless reductionism within the discipline. That fact is further illustrated by the existence and prominence of a new professional society, the Molecular and Cellular Cognition Society (www.molcellcog.org), whose goal is to ‘‘promote the study of the molecular and cellular basis of cognitive function”. The field now includes more than 100 laboratories and 1000 scientists worldwide. Bickle (2003, 2006a, 2006b) provides a metascientific analysis of the nature of reduction at work in this research and contrasts it with more standard notions in the philosophy of science. The analysis is based on a comparison of techniques employed in various paradigmatic experiments reported in top scientific journals. The common technique is to intervene causally at increasingly ‘‘lower” levels of biological organization in animal models and then to track the specific effects of these interventions on behavior in widely accepted experimental protocols for the cognitive phenomenon under investigation (Fig. 1). Interventions increasingly use the tools of molecular biology and genetics to
J. Bickle / Journal of Physiology - Paris 101 (2007) 230–235
Behavior
PSYCHOLOGY (descriptive)
Neuronal Pathways
NEUROANATOMY
231
Neurons CELLULAR NEUROSCIENCE Synapses
Intra-Neuronal Molecular Pathways
MOLECULAR NEUROSCIENCE
Fig. 1. Schematic illustration of the ‘‘intervene molecularly/cellularly and track behaviorally” account of reduction. Dashed arrows represent causal interventions in experimental animals; the solid arrow represents the phenomena with which the effects of these interventions are measured. Psychology is a descriptive discipline, rather than one that offers causal mechanistic explanations.
both enhance and diminish the activities of specific proteins in intra-neuronal signaling. To date, learning and memory have yielded the field’s most spectacular achievements, but results on sensory experience, attention, arousal, and anxiety are emerging. Social cognition, however, seems to present particularly difficult phenomena for ruthless reductionism to address. These phenomena seem to extend beyond the individual brain—and thus well beyond the individual nerve cells that compose them. The causal interactions that produce them seem to involve several brains—not to mention the rest of the bodies that encase them (especially hormonal and immunological features) and the environments embedding those bodies (physical, social, and cultural). At first glance, prospects would seem dim for finding the cellular and molecular basis for these functions. Even for physicalists, the causal interactions here seem inextricably ‘‘higher level”. Is there any hope for ruthlessly reductive neuroscience here? I will argue that there is, on two related grounds. The first ground is conceptual: the ‘‘reductionist’s epiphany” about causes of molecular events in neurons. To the extent that any cognitive function, social or otherwise, depends on a process that we now understand in terms of molecular mechanisms, the only way to causally engage these processes is via other molecular mechanisms. We
would thus know that there must be a molecular basis for that function, presumably discoverable in the way that other molecular mechanisms have been successfully discovered. The second ground is empirical: social recognition memory consolidation is a process that social cognitive functions depend upon and there is now experimental evidence for its molecular mechanism. Combined, these two grounds imply that there must be molecular mechanisms grounding social cognition, accessible to further ruthlessly reductionistic experimental investigation, even if in normal circumstances these mechanisms are triggered initially by distal causes external to the nervous system. Hope reigns for ruthlessly reductive neuroscience. I will close with a brief reply to the ‘‘mouse social cognition” worry that some higher level scientists use to challenge the role of animal models in the experiments I appeal to. 2. Conceptual considerations: The reductionist’s epiphany In a recent review paper, Ferguson et al. (2002) note that ‘‘all social relationships are dependent upon an organism’s ability to remember conspecifics”. They cite reproduction, territorial defense, and the establishment of dominance hierarchies as examples, but this list is easily extended. All features of social interaction, including those of moral psychology in primates, must ‘‘plug into” the mechanisms
232
J. Bickle / Journal of Physiology - Paris 101 (2007) 230–235
of social recognition memory. A crucial component of social recognition memory is consolidation: the conversion of labile, easily disrupted short-term memories of conspecifics into more durable, accessible long-term form. The lower the level of biological organization at which we understand the mechanisms of social recognition memory consolidation (Fig. 1, left side), the lower we will know that there must be specific mechanisms for all the forms of social cognition that depend causally upon it. Call this implication the reductionist’s epiphany. The basic idea is simple. Molecules can only administer a limited number of direct effects and can only be affected by a limited number of direct causes. Molecules respond only to other molecules (or the biochemical and biophysical entities and processes that constitute them). Thus to the extent that we have explained some ‘‘higher level” phenomenon as a sequence (linear or cyclic) of molecular steps, we know that the only way for another ‘‘higher level” process to employ it—to ‘‘plug into” it causally—is via molecular (or lower) mechanisms. There is no other way, short of appealing to spooky, mysterious, nonphysical forces. This paper’s principal argument is therefore straightforward: 1. If a class of cognitive phenomena depends upon some process and the molecular mechanisms of that process have been verified experimentally, then investigating the molecular mechanisms of that class is not only a viable scientific research project, but also one that can assume a kind of methodological priority. 2. Forms of social cognition and behavior depend upon social recognition memory consolidation and the molecular mechanisms of social recognition memory consolidation have been verified experimentally. 3. Therefore research into the molecular mechanisms of social cognition and behavior is not only a viable research project, but also one that can assume methodological priority. Premise 1 is justified by the argument just given for the reductionist’s epiphany. The justification of premise 2 is the topic of the next section. 3. Empirical considerations: The molecular mechanisms of social recognition memory consolidation Establishing the second premise of this paper’s principal argument takes two steps: first, arguing for the dependence of all social behaviors, including complex primate social behaviors, on social recognition memory consolidation; second, describing recent empirical work that provides explicit evidence for the molecular mechanisms of social recognition memory consolidation. The first step is relatively simple. Social behaviors and cognition takes place over time—usually long stretches of time. Social creatures need to not only recognize, but remember and access information tied to particular conspecifics. This is just as true for a wolf pack coordinating an attack on large prey as it
is for a football coach planning for next Sunday’s big game. Social animals need to know who’s who and who’s supposed to do what. And they need to retain and access this information for extended periods of time. This requires consolidating memories of specific individuals from shortterm into long-term form. But how can we best uncover the mechanisms of social recognition memory consolidation in the laboratory? Again Ferguson et al. (2002) describe the favored experimental procedure: ‘‘In the laboratory, social memory can be assessed reliably by measuring the reduction in investigation time of a familiar partner relative to a novel conspecific”. For more than 20 years, experimentalists have employed versions of a protocol first described by Thor and Holloway (1982). A novel juvenile male rodent is placed into the cage of an adult conspecific for a short time (e.g., 2 min), during which the adult will spend some portion of that exposure time displaying stereotypic rodent exploratory behavior. Exploratory time is measured and recorded, and the juvenile is removed from the cage. After a delay period, usually 30 min to 1 h for short-term memory and 24 h for long-term memory, the same juvenile is placed back into the adult’s cage, and again exploratory time by the adult is measured and recorded. The difference (reduction) in exploratory time between the initial and subsequent exposures is taken to be a measure of remembered familiarity of the individual juvenile. Interventions can be introduced at the time of learning and during the consolidation phase to keep intact short-term social recognition memory but block its consolidation into long-term form. Having been used by experimental psychologists now for more than two decades, every potential confound in this protocol has been explored. The experimental work I am about to describe employed the Thor and Holloway behavioral protocol. But it also occurred against a background of what was already known about the molecular mechanisms of late long-term potentiation (LTP) and the experimental verification that these were also the mechanisms of memory consolidation for a variety of hippocampal-dependent memory phenomena.1 Cyclic adenosine monophosphate (cAMP) responsiveelement binding (CREB) proteins, especially two of its isoforms, a and d, are central. These are transcriptional activators that turn on specific gene expression. CREB itself is activated by phosphorylazation by catalytic subunits of cAMP-dependent protein kinase A (PKA), which have translocated to the neuron’s nucleus after being released from regulatory PKA subunits at active synapses. Phosphorylated CREB a and d isoforms turn on gene expression and protein synthesis of both regulatory proteins (like ubiquitin carboxyl-terminal hydrolase, or uch) and other transcriptional activators (like CCAAT enhancer binding protein, or C/EBP), ultimately keeping the synapse
1 My presentation of the scientific details here will be brief. For a more thorough presentation see Bickle (2006a).
J. Bickle / Journal of Physiology - Paris 101 (2007) 230–235
potentiated for days to weeks. Previous work with CREBad mutant mice, which show no expression of CREB a and d isoforms, had demonstrated that these mutants were intact in a variety of short-term memory tasks (delays between stimulus and test 30 min up to 2 h) but significantly impaired on long-term versions of the same tasks (delays of 24–48 h).2 Experimental controls in these studies isolated the defect as one of memory consolidation, rather than sensory, motor, or attentional. Kogan et al. (2000) used CREBad mutant mice in the Thor and Holloway protocol. CREBad mutants were intact (compared to wild-type littermate controls) on short-term (30 min) social recognition memory for the reintroduced juvenile, but were significantly impaired on long-term (24 h) test. There was statistically significant reduction in investigation duration in both mutants and wild-type littermate controls at 30 min, but only for wildtypes at 24 h. (In fact the mean investigative duration at 24 h for the mutants actually increased beyond initial exposure time, although not significantly.) CREBad mutant performance was comparable on both short- and long-term tasks to hippocampus-lesioned wild-type mice and wildtype mice injected systemically with anisomycin (a protein synthesis inhibitor) 30 min prior to initial exposure. According to Kogan et al. (2000), these findings that long-term social memory is dependent on CREB function ‘‘add to the growing evidence that the hippocampus is not limited to processing spatial information” and ‘‘parallels previous findings with other hippocampus-dependent tasks, including the social transmission of food preferences, water maze, and contextual fear conditioning”. Another intriguing result reported in the same study concerned short- and long-term social recognition memory in mice exposed to isolated living conditions. Mice in Silva’s lab are routinely group-housed. To test the effects of social isolation, Kogan et al. (2000) isolated wild-type mice into individual shoebox cages both chronically (for 3 weeks prior to training and testing) and acutely (24 h prior). Both group-housed and chronically isolated mice showed significant reduction in exploratory times with re-introduced juveniles 30 min after initial exposure, indicating intact short-term social recognition memory. But neither chronically nor acutely isolated mice showed significant reduction in exploratory time to the re-introduced juvenile when tested 24 h after initial exposure, indicating deficient long-term social recognition memory consolidation. A comparison of these results with CREBad mutants and anisomycin-treated mice is suggestive. Are availability and normal activity of CREB a and d isoforms the proximal molecular causes of so distal and ‘‘external” an environmental factor as socially isolated living conditions? Obviously we cannot infer this conclusion from this single result. But if CREB can be shown to be expressed in mouse 2
A seminal paper using these mutants in memory research is Bourtchouladze et al. (1994). They have been used in many published studies since then.
233
hippocampus during the consolidation period for social recognition memory, and if intervening into that expression or availability can produce effects on social recognition memory consolidation independent of the animals’ actual living conditions—if, for example, we could produce normal long-term social recognition memory behavior in chronically or acutely isolated mice by boosting CREB availability and activity in the appropriate hippocampal regions—we would have strong evidence for having found an ‘‘internal” molecular mechanism for a specific socialenvironmental cause on a measurable memory behavior. Not only would such a result be scientifically and potentially clinically interesting, but it would also appear to constitute empirical evidence against currently popular strong ‘‘externalist” (or ‘‘embedded”) accounts of mind in philosophy and cognitive science. Whether these speculations pan out experimentally or not, the evidence described in this section supports the second premise of this paper’s principal argument. Most (if not all) forms of social cognition and behavior depend upon social recognition memory consolidation, and the molecular mechanisms of social recognition memory consolidation have been verified experimentally. These mechanisms of phosphorylated CREB a and d transcriptional enhancer isoforms and their gene targets are crucial, leading ultimately to long-term potentiated synapses. With the two premises of this paper’s principal argument now defended, we draw the valid conclusion: Investigating the molecular mechanisms of social cognition and behavior is not only a viable scientific research project, but also one that can now assume a kind of methodological priority. Does this mean that research into ‘‘external” or environmental causal factors of social behavior and cognition should cease? Of course not. Such research constitutes an essential ‘‘leg” in the case for a scientifically viable causal account (see Silva, this volume). But because we know the molecular mechanisms of social recognition memory consolidation, we can fruitfully inquire into other molecular processes that ‘‘plug into” them. In pursuing this research, we will reveal experimentally the molecular mechanisms of forms of social behavior and cognition that require social recognition memory consolidation. Far from being irrelevant or unhelpful in investigating social cognition, the combination of the reductionist’s epiphany with empirical evidence shows that ‘‘ruthlessly reductive” molecular and cellular cognition is center stage in the current endeavor. 4. Is this just ‘‘mouse social cognition? I close this paper by considering a challenge that some philosophers and cognitive scientists raise for the basic methodology employed in molecular and cellular cognition. The field stresses the use of animal models for
234
J. Bickle / Journal of Physiology - Paris 101 (2007) 230–235
cognitive function, and for good reason—the experimental interventions cannot be performed in vivo (ethically, if not practically) on humans, primates, and even some higher mammals. This restriction limits the behavioral tests that can be used to measure cognitive function. But even more worrisome, it raises the issue of cross-species similarity with regard to the hypothesized causal mechanisms. Can we legitimately infer similar mechanisms of, e.g., social recognition memory consolidation from mice to primates? Or is this just ‘‘mouse social cognition” and its molecular mechanisms? Short of relying on experimental genius in developing measures for cognitive function, there is little that can be done about the behavioral limitations of the animal models employed. Experimenters seek behavioral tasks that can be used across species and that intuitively meet the conditions on the cognitive phenomena being investigated. In their choices, molecular and cellular cognitivists tend to defer to the experts—experimental psychologists who are trained in their discipline’s long history of developing behavioral protocols and controlling for possible confounds. The tasks routinely employed in these studies are standard and widely accepted for the phenomena being investigated. The second worry, about molecular mechanisms shared across species, is potentially more damaging. A condition on the reductionist’s epiphany, which supports the first premise of this paper’s principal argument, is our explicit understanding of specific molecular mechanisms for cognitive functions on which other functions depend. Without that, even if (being physicalists) we are confident that such molecular mechanisms must exist, we would not have any knowledge to guide us in experimental investigations. How can we be confident, then, that the mechanisms underlying social cognitive functions revealed by molecular-biological interventions into rodents are shared among higher mammals and primates? Comparative molecular biology across species that consolidate memories justifies our confidence. The molecular biology of memory consolidation didn’t start with rodent studies. Invertebrates provided the first animal models.3 More than thirty years ago Seymour Benzer’s group developed an olfactory shock-avoidance conditioning procedure for the fruit fly Drosophila melanogaster and soon produced the first single-gene learning and memory mutant, dunce. Other training and testing procedures were developed, including several operant procedures. The number of single-gene Drosophila learning and memory mutants is now greater than twenty. A second popular invertebrate preparation for studying cellular and molecular mechanisms of learning and memory has been the marine invertebrate Aplysia californica (a sea slug). A common Aplysia behavioral assay uses its gill withdrawal response. Shortterm gill withdrawal to a neutral stimulus persists for min-
utes after a single sensitizing shock or conditioned–unconditioned stimulus pairing. Five or more shocks spaced appropriately yields long-term sensitization or conditioning that persists for two or more days, up to 2 weeks. A methodologically useful feature of the Aplysia siphon sensory-gill motor circuit is that its components can be extracted and reconstituted in vitro. Short-term facilitation in this preparation can be induced with a single puff (1 lM) of serotonin directly onto the pre-synaptic terminal, and long-term facilitation measurable 24 h later can be induced by five puffs of serotonin delivered over one-and-one-half hours. This preparation enabled detailed investigation of the molecular genetics of late LTP, yielding the information used to build the molecular mouse mutants used in the experiments discussed above. Experimental research has revealed that the same molecular components—cAMP, PKA, and CREB isoforms, plus the same regulatory and effector gene targets (e.g., uch, C/ EBP)—are at work in the consolidation switch from shortto long-term memory across all these species. The only difference is the predominantly pre-synaptic effects in invertebrates (and thus effects on glutamate release) versus postsynaptic effects in mammals (and thus effects on number and conductance affinities of receptor proteins). These comparisons hold despite widely different evolutionary histories (of insects, gastropods, and mammals), sensory modalities, and behavioral repertoires, and even the different forms of memory involved (nondeclarative in the case of invertebrates, declarative in the case of mammals). And the comparisons grow even more significant when we consider the biochemical compositions of the molecules involved. Considering only the CREB homologues discussed here (similar accounts hold for PKA, other CREB isoforms, uch, and C/EBP), biochemical analyses reveal that mammalian CREB a displays 95% amino acid sequence identity with Aplysia ApCREB1a protein in both its C-terminal DNA binding and dimerization domain (bZIP) and its phosphorylation domain (P box). The key phosphorylation sites in the two proteins’ P boxes, the site where freed PKA catalytic subunits induce their effects, are completely conserved between ApCREB1a and mammalian CREB a (Bartsch et al., 1995). Not only do these molecules have similar functions across species; even structurally, they are virtually identical proteins. What these comparisons reflect are general principles of molecular evolution concerning the relatively slow rate of change across species in the ‘‘functionally constrained” regions of genes and their protein products.4 Related principles hold for entire genes and proteins: the amino acid sequences of ‘‘housekeeping” proteins, especially ones that function in basic metabolic processes in all cell types, evolve at much slower rates than do proteins with more specialized cellular functions or limited to specific cell types
3
Again, I will be brief here with scientific details. For a more complete account with extensive references to the primary experimental literature see Bickle (2003, chapter 3).
4 I will only have space to suggest this argument here. The full details are found in Bickle (2003, chapter 3, section 7).
J. Bickle / Journal of Physiology - Paris 101 (2007) 230–235
(Ridley, 2003, chapter 7). These general principles of molecular evolution were already so secure empirically more than two decades ago that Kimura (1983) used them as key premises in his argument for the ‘‘neutral” theory of molecular evolution, a competitor to ‘‘selectionist” accounts. CREB isoforms are just such proteins, occurring and functioning as transcriptional factors in a wide variety of biological tissues. Even single amino acid changes in the functionally constrained regions of these proteins would be lethal, virtually without exception. Hence insecta CREB = gastropod CREB = mammalian CREB, down to the level of amino acid sequences.5 The importance of this fact for the ‘‘rodent social cognition” objection is straightforward. Not only have the molecular mechanisms of memory consolidation been conserved across mammalian evolution, they have been conserved since the evolutionary ancestors we mammals share with the gastropods and insects. When it comes to the molecular mechanisms of cognition, we should expect continuity across species, especially species as closely related evolutionarily as mammals. Even miniscule changes to those components results in dysfunctional neurons. The differences in cognitive complexity across species result from the greater number and connectivities of the similar building blocks. In terms of molecular mechanisms, the building blocks of rodent cognition by and large are the building blocks of primate cognition, humans included. 5. Conclusion Social cognition appears on its face to present problematic phenomena for ruthless mind-brain reductionism. However, a general strategy is to investigate the molecular mechanisms of particular phenomena, like social recognition memory consolidation, that most or all social cognitive functions depend upon. To the extent that we understand the actual molecular mechanisms of these underlying phenomena, the ‘‘reductionist’s epiphany”
5 Why then are CREBad mutants viable? First, their viability depends upon some sophisticated biotechnological tricks (Kogan et al., 2000). Second, in many types of tissue, the transcriptional functions of CREB a and d isoforms can be picked up by cAMP regulatory-element modulating (CREM) proteins—just not completely for those proteins’ roles in memory consolidation.
235
guarantees us that molecular mechanisms for the broader phenomena exist for the finding. And in empirical fact, molecular mechanisms for social recognition memory consolidation have been found, that obtain across widely divergent species. Thus even for causes as distal and external as social interactions, we can be sure that molecular and cellular cognition can find the mechanisms by which such causes are transduced into the currency that real nervous systems trade: the molecular machinery of neural conductance and transmission. This is not based upon just some vague commitment to physicalism about mind, but rather on specific scientific discoveries.
References Bartsch, D., Ghirardi, M., Skehel, P., Karl, K., Herder, S., Chen, M., Bailey, C., Kandel, E., 1995. Aplysia CREB2 represses long-term facilitation: Relief of repression converts transient facilitation into long-term functional and structural change. Cell 83, 979–992. Bickle, J., 2003. Philosophy and Neuroscience: A Ruthlessly Reductive Account. Kluwer, Dordrecht. Bickle, J., 2006a. Reducing mind to molecular pathways: Explicating the reductionism implicit in cellular and molecular neuroscience. Synthese 151, 411–434. Bickle, J., 2006b. Ruthless reductionism in recent neuroscience. IEEE Transactions on Systems, Man, and Cybernetics 36, 134–140. Bourtchouladze, R., Frenguelli, B., Blendy, J., Cioffi, D., Schutz, G., Silva, A., 1994. Deficient long-term memory in mice with a targeted mutation of the cAMP responsive element binding protein. Cell 79, 59–68. Ferguson, J., Young, L., Insel, T., 2002. The neuroendocrine basis of social recognition. Frontiers in Neuroendocrinology 23, 200–224. Kandel, E., Schwartz, J., Jessell, T., 2001. Principles of Neural Science, fourth ed. McGraw-Hill, New York. Kimura, M., 1983. The Neutral Theory of Molecular Evolution. Cambridge University Press, Cambridge. Kogan, J., Frankland, P., Silva, A., 2000. Long-term memory underlying hippocampus-dependent social recognition in mice. Hippocampus 10, 47–56. Ridley, M., 2003. Evolution, third ed. Blackwell, New York. Thor, D., Holloway, W., 1982. Social memory of the male laboratory rat. Journal of Comparative and Physiological Psychology 96, 1000–1006.
Journal of Physiology - Paris 101 (2007) 230–235 www.elsevier.com/locate/jphysparis
Ruthless reductionism and social cognition John Bickle Department of Philosophy and Neuroscience Graduate Program, University of Cincinnati, Cincinnati, OH, USA
Abstract Social cognition appears to present phenomena that ‘‘ruthlessly reductive” molecular and cellular neuroscience cannot fruitfully investigate or explain. This is because the causes of such phenomena are distal and external not only to the molecular machinery of individual neurons, but to individual brains. However, the ‘‘reductionist’s epiphany” insists that to the extent that we understand the specific molecular mechanisms that underlie phenomena upon which most or all social cognition depends, we can be sure that molecular mechanisms for the broader phenomena can be found using standard experimental methods from molecular and cellular cognition. Furthermore, social recognition memory consolidation is required for virtually all types of social cognition, and its specific molecular mechanisms have now been uncovered experimentally. These same molecular mechanisms obtain across a wide variety of divergent species (from invertebrates to vertebrates). Thus we can expect to find the molecular mechanisms of the broader social cognitive functions that must ‘‘plug into” these specific molecular mechanisms, despite these functions’ typically distal, external initial causes. This conclusion rests on explicit scientific facts, not just on some vague philosophical commitment to physicalism about mind. Ó 2008 Elsevier Ltd. All rights reserved. Keywords: Reductionist’s epiphany; Social recognition memory consolidation; cAMP-PKA-CREB pathway; CREBad mutant mice; Molecular evolution
1. Introduction: Ruthless reductionism in recent neuroscience ‘‘Ruthless reductionism” informs key branches of contemporary neuroscience. Its features are nicely illustrated in a passage from the introductory chapter of Kandel, Schwartz, and Jessell’s state-of-the-art textbook, Principles of Neural Science: This book . . . describes how neural science is attempting to link molecules to mind—how proteins responsible for the activities of individual nerve cells are related to the complexity of neural processes. Today it is possible to link the molecular dynamics of individual nerve cells to representations of perceptual and motor acts in the brain and to relate these internal mechanisms to observable behavior. (2001, 3–4) This passage nicely illustrates that the explanatory target of these investigations is mind, not some ersatz laboratory stand-in. The explanations and experimental manipulations
E-mail address: [email protected] 0928-4257/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.jphysparis.2007.11.009
have now reached into molecular mechanisms internal to individual neurons. That this passage occurs in one of neuroscience’s principal current textbooks indicates the centrality of ruthless reductionism within the discipline. That fact is further illustrated by the existence and prominence of a new professional society, the Molecular and Cellular Cognition Society (www.molcellcog.org), whose goal is to ‘‘promote the study of the molecular and cellular basis of cognitive function”. The field now includes more than 100 laboratories and 1000 scientists worldwide. Bickle (2003, 2006a, 2006b) provides a metascientific analysis of the nature of reduction at work in this research and contrasts it with more standard notions in the philosophy of science. The analysis is based on a comparison of techniques employed in various paradigmatic experiments reported in top scientific journals. The common technique is to intervene causally at increasingly ‘‘lower” levels of biological organization in animal models and then to track the specific effects of these interventions on behavior in widely accepted experimental protocols for the cognitive phenomenon under investigation (Fig. 1). Interventions increasingly use the tools of molecular biology and genetics to
J. Bickle / Journal of Physiology - Paris 101 (2007) 230–235
Behavior
PSYCHOLOGY (descriptive)
Neuronal Pathways
NEUROANATOMY
231
Neurons CELLULAR NEUROSCIENCE Synapses
Intra-Neuronal Molecular Pathways
MOLECULAR NEUROSCIENCE
Fig. 1. Schematic illustration of the ‘‘intervene molecularly/cellularly and track behaviorally” account of reduction. Dashed arrows represent causal interventions in experimental animals; the solid arrow represents the phenomena with which the effects of these interventions are measured. Psychology is a descriptive discipline, rather than one that offers causal mechanistic explanations.
both enhance and diminish the activities of specific proteins in intra-neuronal signaling. To date, learning and memory have yielded the field’s most spectacular achievements, but results on sensory experience, attention, arousal, and anxiety are emerging. Social cognition, however, seems to present particularly difficult phenomena for ruthless reductionism to address. These phenomena seem to extend beyond the individual brain—and thus well beyond the individual nerve cells that compose them. The causal interactions that produce them seem to involve several brains—not to mention the rest of the bodies that encase them (especially hormonal and immunological features) and the environments embedding those bodies (physical, social, and cultural). At first glance, prospects would seem dim for finding the cellular and molecular basis for these functions. Even for physicalists, the causal interactions here seem inextricably ‘‘higher level”. Is there any hope for ruthlessly reductive neuroscience here? I will argue that there is, on two related grounds. The first ground is conceptual: the ‘‘reductionist’s epiphany” about causes of molecular events in neurons. To the extent that any cognitive function, social or otherwise, depends on a process that we now understand in terms of molecular mechanisms, the only way to causally engage these processes is via other molecular mechanisms. We
would thus know that there must be a molecular basis for that function, presumably discoverable in the way that other molecular mechanisms have been successfully discovered. The second ground is empirical: social recognition memory consolidation is a process that social cognitive functions depend upon and there is now experimental evidence for its molecular mechanism. Combined, these two grounds imply that there must be molecular mechanisms grounding social cognition, accessible to further ruthlessly reductionistic experimental investigation, even if in normal circumstances these mechanisms are triggered initially by distal causes external to the nervous system. Hope reigns for ruthlessly reductive neuroscience. I will close with a brief reply to the ‘‘mouse social cognition” worry that some higher level scientists use to challenge the role of animal models in the experiments I appeal to. 2. Conceptual considerations: The reductionist’s epiphany In a recent review paper, Ferguson et al. (2002) note that ‘‘all social relationships are dependent upon an organism’s ability to remember conspecifics”. They cite reproduction, territorial defense, and the establishment of dominance hierarchies as examples, but this list is easily extended. All features of social interaction, including those of moral psychology in primates, must ‘‘plug into” the mechanisms
232
J. Bickle / Journal of Physiology - Paris 101 (2007) 230–235
of social recognition memory. A crucial component of social recognition memory is consolidation: the conversion of labile, easily disrupted short-term memories of conspecifics into more durable, accessible long-term form. The lower the level of biological organization at which we understand the mechanisms of social recognition memory consolidation (Fig. 1, left side), the lower we will know that there must be specific mechanisms for all the forms of social cognition that depend causally upon it. Call this implication the reductionist’s epiphany. The basic idea is simple. Molecules can only administer a limited number of direct effects and can only be affected by a limited number of direct causes. Molecules respond only to other molecules (or the biochemical and biophysical entities and processes that constitute them). Thus to the extent that we have explained some ‘‘higher level” phenomenon as a sequence (linear or cyclic) of molecular steps, we know that the only way for another ‘‘higher level” process to employ it—to ‘‘plug into” it causally—is via molecular (or lower) mechanisms. There is no other way, short of appealing to spooky, mysterious, nonphysical forces. This paper’s principal argument is therefore straightforward: 1. If a class of cognitive phenomena depends upon some process and the molecular mechanisms of that process have been verified experimentally, then investigating the molecular mechanisms of that class is not only a viable scientific research project, but also one that can assume a kind of methodological priority. 2. Forms of social cognition and behavior depend upon social recognition memory consolidation and the molecular mechanisms of social recognition memory consolidation have been verified experimentally. 3. Therefore research into the molecular mechanisms of social cognition and behavior is not only a viable research project, but also one that can assume methodological priority. Premise 1 is justified by the argument just given for the reductionist’s epiphany. The justification of premise 2 is the topic of the next section. 3. Empirical considerations: The molecular mechanisms of social recognition memory consolidation Establishing the second premise of this paper’s principal argument takes two steps: first, arguing for the dependence of all social behaviors, including complex primate social behaviors, on social recognition memory consolidation; second, describing recent empirical work that provides explicit evidence for the molecular mechanisms of social recognition memory consolidation. The first step is relatively simple. Social behaviors and cognition takes place over time—usually long stretches of time. Social creatures need to not only recognize, but remember and access information tied to particular conspecifics. This is just as true for a wolf pack coordinating an attack on large prey as it
is for a football coach planning for next Sunday’s big game. Social animals need to know who’s who and who’s supposed to do what. And they need to retain and access this information for extended periods of time. This requires consolidating memories of specific individuals from shortterm into long-term form. But how can we best uncover the mechanisms of social recognition memory consolidation in the laboratory? Again Ferguson et al. (2002) describe the favored experimental procedure: ‘‘In the laboratory, social memory can be assessed reliably by measuring the reduction in investigation time of a familiar partner relative to a novel conspecific”. For more than 20 years, experimentalists have employed versions of a protocol first described by Thor and Holloway (1982). A novel juvenile male rodent is placed into the cage of an adult conspecific for a short time (e.g., 2 min), during which the adult will spend some portion of that exposure time displaying stereotypic rodent exploratory behavior. Exploratory time is measured and recorded, and the juvenile is removed from the cage. After a delay period, usually 30 min to 1 h for short-term memory and 24 h for long-term memory, the same juvenile is placed back into the adult’s cage, and again exploratory time by the adult is measured and recorded. The difference (reduction) in exploratory time between the initial and subsequent exposures is taken to be a measure of remembered familiarity of the individual juvenile. Interventions can be introduced at the time of learning and during the consolidation phase to keep intact short-term social recognition memory but block its consolidation into long-term form. Having been used by experimental psychologists now for more than two decades, every potential confound in this protocol has been explored. The experimental work I am about to describe employed the Thor and Holloway behavioral protocol. But it also occurred against a background of what was already known about the molecular mechanisms of late long-term potentiation (LTP) and the experimental verification that these were also the mechanisms of memory consolidation for a variety of hippocampal-dependent memory phenomena.1 Cyclic adenosine monophosphate (cAMP) responsiveelement binding (CREB) proteins, especially two of its isoforms, a and d, are central. These are transcriptional activators that turn on specific gene expression. CREB itself is activated by phosphorylazation by catalytic subunits of cAMP-dependent protein kinase A (PKA), which have translocated to the neuron’s nucleus after being released from regulatory PKA subunits at active synapses. Phosphorylated CREB a and d isoforms turn on gene expression and protein synthesis of both regulatory proteins (like ubiquitin carboxyl-terminal hydrolase, or uch) and other transcriptional activators (like CCAAT enhancer binding protein, or C/EBP), ultimately keeping the synapse
1 My presentation of the scientific details here will be brief. For a more thorough presentation see Bickle (2006a).
J. Bickle / Journal of Physiology - Paris 101 (2007) 230–235
potentiated for days to weeks. Previous work with CREBad mutant mice, which show no expression of CREB a and d isoforms, had demonstrated that these mutants were intact in a variety of short-term memory tasks (delays between stimulus and test 30 min up to 2 h) but significantly impaired on long-term versions of the same tasks (delays of 24–48 h).2 Experimental controls in these studies isolated the defect as one of memory consolidation, rather than sensory, motor, or attentional. Kogan et al. (2000) used CREBad mutant mice in the Thor and Holloway protocol. CREBad mutants were intact (compared to wild-type littermate controls) on short-term (30 min) social recognition memory for the reintroduced juvenile, but were significantly impaired on long-term (24 h) test. There was statistically significant reduction in investigation duration in both mutants and wild-type littermate controls at 30 min, but only for wildtypes at 24 h. (In fact the mean investigative duration at 24 h for the mutants actually increased beyond initial exposure time, although not significantly.) CREBad mutant performance was comparable on both short- and long-term tasks to hippocampus-lesioned wild-type mice and wildtype mice injected systemically with anisomycin (a protein synthesis inhibitor) 30 min prior to initial exposure. According to Kogan et al. (2000), these findings that long-term social memory is dependent on CREB function ‘‘add to the growing evidence that the hippocampus is not limited to processing spatial information” and ‘‘parallels previous findings with other hippocampus-dependent tasks, including the social transmission of food preferences, water maze, and contextual fear conditioning”. Another intriguing result reported in the same study concerned short- and long-term social recognition memory in mice exposed to isolated living conditions. Mice in Silva’s lab are routinely group-housed. To test the effects of social isolation, Kogan et al. (2000) isolated wild-type mice into individual shoebox cages both chronically (for 3 weeks prior to training and testing) and acutely (24 h prior). Both group-housed and chronically isolated mice showed significant reduction in exploratory times with re-introduced juveniles 30 min after initial exposure, indicating intact short-term social recognition memory. But neither chronically nor acutely isolated mice showed significant reduction in exploratory time to the re-introduced juvenile when tested 24 h after initial exposure, indicating deficient long-term social recognition memory consolidation. A comparison of these results with CREBad mutants and anisomycin-treated mice is suggestive. Are availability and normal activity of CREB a and d isoforms the proximal molecular causes of so distal and ‘‘external” an environmental factor as socially isolated living conditions? Obviously we cannot infer this conclusion from this single result. But if CREB can be shown to be expressed in mouse 2
A seminal paper using these mutants in memory research is Bourtchouladze et al. (1994). They have been used in many published studies since then.
233
hippocampus during the consolidation period for social recognition memory, and if intervening into that expression or availability can produce effects on social recognition memory consolidation independent of the animals’ actual living conditions—if, for example, we could produce normal long-term social recognition memory behavior in chronically or acutely isolated mice by boosting CREB availability and activity in the appropriate hippocampal regions—we would have strong evidence for having found an ‘‘internal” molecular mechanism for a specific socialenvironmental cause on a measurable memory behavior. Not only would such a result be scientifically and potentially clinically interesting, but it would also appear to constitute empirical evidence against currently popular strong ‘‘externalist” (or ‘‘embedded”) accounts of mind in philosophy and cognitive science. Whether these speculations pan out experimentally or not, the evidence described in this section supports the second premise of this paper’s principal argument. Most (if not all) forms of social cognition and behavior depend upon social recognition memory consolidation, and the molecular mechanisms of social recognition memory consolidation have been verified experimentally. These mechanisms of phosphorylated CREB a and d transcriptional enhancer isoforms and their gene targets are crucial, leading ultimately to long-term potentiated synapses. With the two premises of this paper’s principal argument now defended, we draw the valid conclusion: Investigating the molecular mechanisms of social cognition and behavior is not only a viable scientific research project, but also one that can now assume a kind of methodological priority. Does this mean that research into ‘‘external” or environmental causal factors of social behavior and cognition should cease? Of course not. Such research constitutes an essential ‘‘leg” in the case for a scientifically viable causal account (see Silva, this volume). But because we know the molecular mechanisms of social recognition memory consolidation, we can fruitfully inquire into other molecular processes that ‘‘plug into” them. In pursuing this research, we will reveal experimentally the molecular mechanisms of forms of social behavior and cognition that require social recognition memory consolidation. Far from being irrelevant or unhelpful in investigating social cognition, the combination of the reductionist’s epiphany with empirical evidence shows that ‘‘ruthlessly reductive” molecular and cellular cognition is center stage in the current endeavor. 4. Is this just ‘‘mouse social cognition? I close this paper by considering a challenge that some philosophers and cognitive scientists raise for the basic methodology employed in molecular and cellular cognition. The field stresses the use of animal models for
234
J. Bickle / Journal of Physiology - Paris 101 (2007) 230–235
cognitive function, and for good reason—the experimental interventions cannot be performed in vivo (ethically, if not practically) on humans, primates, and even some higher mammals. This restriction limits the behavioral tests that can be used to measure cognitive function. But even more worrisome, it raises the issue of cross-species similarity with regard to the hypothesized causal mechanisms. Can we legitimately infer similar mechanisms of, e.g., social recognition memory consolidation from mice to primates? Or is this just ‘‘mouse social cognition” and its molecular mechanisms? Short of relying on experimental genius in developing measures for cognitive function, there is little that can be done about the behavioral limitations of the animal models employed. Experimenters seek behavioral tasks that can be used across species and that intuitively meet the conditions on the cognitive phenomena being investigated. In their choices, molecular and cellular cognitivists tend to defer to the experts—experimental psychologists who are trained in their discipline’s long history of developing behavioral protocols and controlling for possible confounds. The tasks routinely employed in these studies are standard and widely accepted for the phenomena being investigated. The second worry, about molecular mechanisms shared across species, is potentially more damaging. A condition on the reductionist’s epiphany, which supports the first premise of this paper’s principal argument, is our explicit understanding of specific molecular mechanisms for cognitive functions on which other functions depend. Without that, even if (being physicalists) we are confident that such molecular mechanisms must exist, we would not have any knowledge to guide us in experimental investigations. How can we be confident, then, that the mechanisms underlying social cognitive functions revealed by molecular-biological interventions into rodents are shared among higher mammals and primates? Comparative molecular biology across species that consolidate memories justifies our confidence. The molecular biology of memory consolidation didn’t start with rodent studies. Invertebrates provided the first animal models.3 More than thirty years ago Seymour Benzer’s group developed an olfactory shock-avoidance conditioning procedure for the fruit fly Drosophila melanogaster and soon produced the first single-gene learning and memory mutant, dunce. Other training and testing procedures were developed, including several operant procedures. The number of single-gene Drosophila learning and memory mutants is now greater than twenty. A second popular invertebrate preparation for studying cellular and molecular mechanisms of learning and memory has been the marine invertebrate Aplysia californica (a sea slug). A common Aplysia behavioral assay uses its gill withdrawal response. Shortterm gill withdrawal to a neutral stimulus persists for min-
utes after a single sensitizing shock or conditioned–unconditioned stimulus pairing. Five or more shocks spaced appropriately yields long-term sensitization or conditioning that persists for two or more days, up to 2 weeks. A methodologically useful feature of the Aplysia siphon sensory-gill motor circuit is that its components can be extracted and reconstituted in vitro. Short-term facilitation in this preparation can be induced with a single puff (1 lM) of serotonin directly onto the pre-synaptic terminal, and long-term facilitation measurable 24 h later can be induced by five puffs of serotonin delivered over one-and-one-half hours. This preparation enabled detailed investigation of the molecular genetics of late LTP, yielding the information used to build the molecular mouse mutants used in the experiments discussed above. Experimental research has revealed that the same molecular components—cAMP, PKA, and CREB isoforms, plus the same regulatory and effector gene targets (e.g., uch, C/ EBP)—are at work in the consolidation switch from shortto long-term memory across all these species. The only difference is the predominantly pre-synaptic effects in invertebrates (and thus effects on glutamate release) versus postsynaptic effects in mammals (and thus effects on number and conductance affinities of receptor proteins). These comparisons hold despite widely different evolutionary histories (of insects, gastropods, and mammals), sensory modalities, and behavioral repertoires, and even the different forms of memory involved (nondeclarative in the case of invertebrates, declarative in the case of mammals). And the comparisons grow even more significant when we consider the biochemical compositions of the molecules involved. Considering only the CREB homologues discussed here (similar accounts hold for PKA, other CREB isoforms, uch, and C/EBP), biochemical analyses reveal that mammalian CREB a displays 95% amino acid sequence identity with Aplysia ApCREB1a protein in both its C-terminal DNA binding and dimerization domain (bZIP) and its phosphorylation domain (P box). The key phosphorylation sites in the two proteins’ P boxes, the site where freed PKA catalytic subunits induce their effects, are completely conserved between ApCREB1a and mammalian CREB a (Bartsch et al., 1995). Not only do these molecules have similar functions across species; even structurally, they are virtually identical proteins. What these comparisons reflect are general principles of molecular evolution concerning the relatively slow rate of change across species in the ‘‘functionally constrained” regions of genes and their protein products.4 Related principles hold for entire genes and proteins: the amino acid sequences of ‘‘housekeeping” proteins, especially ones that function in basic metabolic processes in all cell types, evolve at much slower rates than do proteins with more specialized cellular functions or limited to specific cell types
3
Again, I will be brief here with scientific details. For a more complete account with extensive references to the primary experimental literature see Bickle (2003, chapter 3).
4 I will only have space to suggest this argument here. The full details are found in Bickle (2003, chapter 3, section 7).
J. Bickle / Journal of Physiology - Paris 101 (2007) 230–235
(Ridley, 2003, chapter 7). These general principles of molecular evolution were already so secure empirically more than two decades ago that Kimura (1983) used them as key premises in his argument for the ‘‘neutral” theory of molecular evolution, a competitor to ‘‘selectionist” accounts. CREB isoforms are just such proteins, occurring and functioning as transcriptional factors in a wide variety of biological tissues. Even single amino acid changes in the functionally constrained regions of these proteins would be lethal, virtually without exception. Hence insecta CREB = gastropod CREB = mammalian CREB, down to the level of amino acid sequences.5 The importance of this fact for the ‘‘rodent social cognition” objection is straightforward. Not only have the molecular mechanisms of memory consolidation been conserved across mammalian evolution, they have been conserved since the evolutionary ancestors we mammals share with the gastropods and insects. When it comes to the molecular mechanisms of cognition, we should expect continuity across species, especially species as closely related evolutionarily as mammals. Even miniscule changes to those components results in dysfunctional neurons. The differences in cognitive complexity across species result from the greater number and connectivities of the similar building blocks. In terms of molecular mechanisms, the building blocks of rodent cognition by and large are the building blocks of primate cognition, humans included. 5. Conclusion Social cognition appears on its face to present problematic phenomena for ruthless mind-brain reductionism. However, a general strategy is to investigate the molecular mechanisms of particular phenomena, like social recognition memory consolidation, that most or all social cognitive functions depend upon. To the extent that we understand the actual molecular mechanisms of these underlying phenomena, the ‘‘reductionist’s epiphany”
5 Why then are CREBad mutants viable? First, their viability depends upon some sophisticated biotechnological tricks (Kogan et al., 2000). Second, in many types of tissue, the transcriptional functions of CREB a and d isoforms can be picked up by cAMP regulatory-element modulating (CREM) proteins—just not completely for those proteins’ roles in memory consolidation.
235
guarantees us that molecular mechanisms for the broader phenomena exist for the finding. And in empirical fact, molecular mechanisms for social recognition memory consolidation have been found, that obtain across widely divergent species. Thus even for causes as distal and external as social interactions, we can be sure that molecular and cellular cognition can find the mechanisms by which such causes are transduced into the currency that real nervous systems trade: the molecular machinery of neural conductance and transmission. This is not based upon just some vague commitment to physicalism about mind, but rather on specific scientific discoveries.
References Bartsch, D., Ghirardi, M., Skehel, P., Karl, K., Herder, S., Chen, M., Bailey, C., Kandel, E., 1995. Aplysia CREB2 represses long-term facilitation: Relief of repression converts transient facilitation into long-term functional and structural change. Cell 83, 979–992. Bickle, J., 2003. Philosophy and Neuroscience: A Ruthlessly Reductive Account. Kluwer, Dordrecht. Bickle, J., 2006a. Reducing mind to molecular pathways: Explicating the reductionism implicit in cellular and molecular neuroscience. Synthese 151, 411–434. Bickle, J., 2006b. Ruthless reductionism in recent neuroscience. IEEE Transactions on Systems, Man, and Cybernetics 36, 134–140. Bourtchouladze, R., Frenguelli, B., Blendy, J., Cioffi, D., Schutz, G., Silva, A., 1994. Deficient long-term memory in mice with a targeted mutation of the cAMP responsive element binding protein. Cell 79, 59–68. Ferguson, J., Young, L., Insel, T., 2002. The neuroendocrine basis of social recognition. Frontiers in Neuroendocrinology 23, 200–224. Kandel, E., Schwartz, J., Jessell, T., 2001. Principles of Neural Science, fourth ed. McGraw-Hill, New York. Kimura, M., 1983. The Neutral Theory of Molecular Evolution. Cambridge University Press, Cambridge. Kogan, J., Frankland, P., Silva, A., 2000. Long-term memory underlying hippocampus-dependent social recognition in mice. Hippocampus 10, 47–56. Ridley, M., 2003. Evolution, third ed. Blackwell, New York. Thor, D., Holloway, W., 1982. Social memory of the male laboratory rat. Journal of Comparative and Physiological Psychology 96, 1000–1006.