Elizabeth Spelke's (Reference Spelke2022) What Babies Know is scholarly, erudite, and often insightful; it is an intriguing, thought-provoking book packed with research results. It begins with an interesting question: “what do human infants know … when their learning begins?” (p. xv). However, no clear definition of “learning” is offered, an omission that proves problematic. Learning entails functional changes that result from experiences, and because embryonic tissues – including ectodermal cells that become the first neurons – are functionally changed by the contexts they experience (Spemann & Mangold, Reference Spemann and Mangold1924/2001), “learning” arguably begins before infants know anything at all. Although there is value in this book's collation of experimental data, the theoretical scaffolding that serves as the work's glue is nondevelopmental and outdated. Spelke's efforts to build a theory on the notion of “core” knowledge have done her empirical work a disservice by situating that work in a nondevelopmental theoretical framework.
Spelke's conceptualization of certain cognitive systems as “core” suggests that she is in thrall to an old Weltanschauung that sees some capabilities as inevitable outcomes of prenatal development, capabilities nineteenth-century theorists would have attributed to “nature” rather than “nurture” (Moore, Reference Moore2001, Reference Moore and Kampourakis2013). But this dichotomous way of thinking about phenotype origins has been rendered obsolete by the work of developmental psychobiologists (Gottlieb, Reference Gottlieb2007; Michel & Moore, Reference Michel and Moore1995) and molecular (Lewontin, Reference Lewontin2000; Strohman, Reference Strohman2003), physiological (Noble & Noble, Reference Noble and Noble2023), and developmental (Gilbert & Epel, Reference Gilbert and Epel2015) biologists. All have conclusively established that phenotypes are emergent products of a probabilistic, multifactorial, and context-dependent developmental process that depends on the interaction and coaction of genetic and nongenetic factors. (Nongenetic factors reside in multiple places: In the environment outside the body and both inside and outside of cells but still inside the body.) The consensus of developmental systems theorists is that cognitive systems emerge from complex, dynamic interactions between – and coactions of – these genetic and nongenetic factors, where emergence is likely to be characterized by reorganization of component systems during development. Clearly, this developmental systems perspective differs radically from the more predeterministic view held by nativists like Spelke, a view that overlooks the need for investigations of the mechanisms underlying the developmental emergence of early-appearing cognitive skills.
The finding that some competencies are present at birth carries great significance for Spelke because it suggests to her that these competencies are not learned. But learning is only one component of experience, and starting with developmental processes that commence at conception, any experiential factor can potentially have profound effects on the developmental emergence of a cognitive/behavioral skill (Gottlieb, Reference Gottlieb1991; Lehrman, Reference Lehrman1953). Indeed, several experience-dependent cognitive competencies have been detected in fetuses, as Spelke discusses in chapter 9; such findings are consistent with the understanding that development is both a continuous process beginning at conception and one that normally entails reorganization.
Furthermore, some species-typical competencies are not present at birth but are no less foundational for normal functioning. Indeed, the timing of appearance of a cognitive competence need not bear any relation to how foundational it is. Therefore, developmental scientists' principal aim should not be to identify cognitive functions present at birth in order to declare them “core” and fundamental to all that emerge later. Rather, once particular functions are discovered, our job is to explicate the processes underlying their emergence in development regardless of when they emerge.
One entrenched, old idea is that there are two different processes responsible for phenotypes, one that relies on experiences and one that yields experience-independent “evolved behaviors.” This idea is known as the “phylogeny fallacy” (Lickliter & Berry, Reference Lickliter and Berry1990) simply because phenotype emergence in each generation is the product of development, and all development involves both genetic and experiential factors. To be sure, Spelke acknowledges that so-called “innate” cognitive processes must develop prenatally and that the experiences that give rise to them might involve activity generated in “subcortical or older cortical regions…that propagates to the plastic [neocortex]” (p. 195). But what brings about such prenatal activity and why should the resultant cognitive processes be considered any more “core” than earlier- or later-emerging processes? After all, the development of every higher cognitive process reflects neocortical cells' experiences with incoming stimulation. Consequently, a developmental analysis must a priori consider all stimulation as potentially crucial unless its role has been empirically ruled out. Such analyses do not need to distinguish between stimulation that arises outside of the neocortex but still within the brain (e.g., in a subcortical region) and stimulation that arises outside of the brain entirely (e.g., circulating hormones in a fetus's body, auditory stimuli that flood a fetus's brain with neurotransmitters, or a newborn's first whiff of its mother). Given this, Spelke's categorical distinction between different kinds of stimuli can be considered arbitrary and offers few insights into the emergence of cognitive skills.
At the heart of Spelke's theoretical argument are these statements: “the core systems … have been shaped by hundreds of millions of years of cognitive evolution. Some core systems are shared by animals as remotely related to us as fish, and aspects of these systems are shared by flies and worms” (p. xx). For example, “the place system is innate. Cognitive and brain scientists have studied the mechanisms and processes by which place representations arise in infant minds, through research on animals who have been reared under systematically controlled conditions” (p. 139). These statements reflect at least three significant misunderstandings.
First, every cognitive system in extant organisms reflects millions of years of evolution and relies on subsystems that have survived natural selection (e.g., functioning neurons, sodium–potassium pumps, etc.). Phenotypes with long evolutionary histories should not be considered any more “core” – in the sense of “atomic” – than more recently evolved phenotypes that contribute to organisms' survival. All cognitive systems are built from smaller components that have long evolutionary histories.
Second, Spelke writes that core systems “have been shaped by…evolution,” a claim that reflects a misunderstanding common among psychologists. This idea conflates the dynamics of populations, which can be influenced by natural selection, with the dynamics of individuals, which cannot (Witherington, Lickliter, & Moore, Reference Witherington, Lickliter and Moore2023). Natural selection does not have the creative power to “shape” individuals' phenotypes (Sober, Reference Sober1984). Although natural selection can, over time, affect the distribution of phenotypes in a population, individuals' phenotypes always reflect the developmental process that is of paramount concern to developmental scientists (Moore, Reference Moore2008).
Finally, investigating how representations arise in human infant minds by doing experiments on nonhuman animals is fraught, because homology is not identity. Bird wings are homologous with human arms, but the bones in bird wings are different in key respects from the bones in our arms (Dumont, Reference Dumont2010). Thus, Spelke is on shaky ground when she claims that the human navigational system is “core” because other animals possess such a system at birth.
Spelke is correct that her “core” systems support children's later learning, and her claim that it is advantageous to develop the capacity for abstraction early in life is insightful; abstraction need not await arrival at some developmental pinnacle. Still, there is no reason to argue that experiences are unnecessary as we develop the ability to abstract information. Although Spelke argues that infants “are predisposed to learn to … categorize objects by using shape descriptions that capture the characteristic forms of plants and animals” (p. 202), there is strong experimental evidence that some of these forms are experience-dependent. For example, the basic categories of faces (e.g., same-race vs. other-race) and language (native vs. nonnative) are acquired through perceptual narrowing during infancy (Lewkowicz & Ghazanfar, Reference Lewkowicz and Ghazanfar2009). Initially, these forms are so broadly specified that newborns do not distinguish human from nonhuman faces (Di Giorgio, Leo, Pascalis, & Simion, Reference Di Giorgio, Leo, Pascalis and Simion2012) and integrate nonhuman faces with temporally synchronized tones (Lewkowicz, Leo, & Simion, Reference Lewkowicz, Leo and Simion2010). These findings challenge Spelke's view that infants are predisposed to recognize “people as social beings” (p. 301). Clearly, that level of cognitive specificity emerges after months of everyday postnatal experience.
Is it worth knowing that a particular ability is functional at birth? Absolutely. But that does not tell us (1) how the ability develops, (2) how important it is, (3) that it develops independently of contextual factors, or (4) that it is any more “core” than later-appearing abilities. Abilities that are important to normal human functioning appear at various times in development: The ability to hear and to suck is present in utero, the ability to navigate is present at birth, and the ability to walk is present only after about a year of postnatal development. There is little about appearing at birth that makes a characteristic special. Furthermore, it is not yet possible to do fair tests of human newborns' competencies in most of Spelke's “core” domains. Consequently, Spelke draws many of her conclusions from studies of older infants, toddlers, or young children. The problem is that all these studies reflect postnatal experiences and therefore do not convincingly demonstrate that these competencies are “innate” in humans.
When Spelke's arguments invoke the commonalities we share with fish, she claims that it is “the same cognitive system, inherited from a distant common ancestor, [that] underlies their performance” (p. 129). However, vertebrates and other complex organisms inherit only developmental resources (including genes used in protein production), resources that we use to build our phenotypes anew in each generation via probabilistic developmental processes; we cannot inherit full blown phenotypes (Lewkowicz, Reference Lewkowicz2011; Moore & Lickliter, Reference Moore and Lickliter2023). In discussing the navigational abilities of fish, Spelke supports her arguments with the results of experiments that deny fish certain experiences. These are the same sorts of “isolation” experiments that Lehrman (Reference Lehrman1953) warned can only reveal if withheld environmental factors are probably not directly involved in the development of a behavior; isolation experiments do not license the conclusion that the behavior under investigation is innate. Spelke's conclusion that such experiments reveal innateness only stymies true developmental analysis.
Spelke believes “the core systems provide the foundations for the abstract concepts at the center of all our explicit knowledge” (p. xix). Perhaps. Nevertheless, this does not free developmental scientists from the responsibility of investigating the developmental emergence of such systems. Of course, it is not Spelke's personal responsibility to do so. Nonetheless, it is one thing to choose a starting point and try to explain what develops from that point, but it is quite another to imply that the developmental processes responsible for bringing about that starting point do not warrant study. Spelke effectively does the latter when she labels these competencies “core.”
The renowned Swiss professor of anatomy and physiology Wilhelm His wrote in 1888:
The single word “heredity” cannot dispense science from the duty of making every possible inquiry into the mechanism of organic growth… . To think that heredity will build organic beings without mechanical means is a piece of unscientific mysticism… . A direct explanation … [of the emergence of phenotypes] can only come from the immediate study of the different phases of individual development. (p. 295)
His highlighted the critical role of developmental analysis, regardless of whether the phenotype has a long or short evolutionary history and regardless of when in ontogeny the phenotype appears. Accordingly, we can reasonably ask nativists why we should explore how children accomplish tasks that confront them at 1 year of age (as Spelke will do in How children learn), but merely accept earlier-appearing competencies as somehow “core” and consequently not requiring developmental analysis. Scientists interested in development should resist the notion of “core” competencies because it short-circuits developmental analysis, leaving us ignorant of the factors that lead to the emergence of these competencies in the first place. Because knowledge of these factors could very well be important for understanding why adult-like competence in domains such as social communication fail to develop (as in autism), understanding their developmental origins should be a central goal of our discipline.
You have
Access
Elizabeth Spelke's (Reference Spelke2022) What Babies Know is scholarly, erudite, and often insightful; it is an intriguing, thought-provoking book packed with research results. It begins with an interesting question: “what do human infants know … when their learning begins?” (p. xv). However, no clear definition of “learning” is offered, an omission that proves problematic. Learning entails functional changes that result from experiences, and because embryonic tissues – including ectodermal cells that become the first neurons – are functionally changed by the contexts they experience (Spemann & Mangold, Reference Spemann and Mangold1924/2001), “learning” arguably begins before infants know anything at all. Although there is value in this book's collation of experimental data, the theoretical scaffolding that serves as the work's glue is nondevelopmental and outdated. Spelke's efforts to build a theory on the notion of “core” knowledge have done her empirical work a disservice by situating that work in a nondevelopmental theoretical framework.
Spelke's conceptualization of certain cognitive systems as “core” suggests that she is in thrall to an old Weltanschauung that sees some capabilities as inevitable outcomes of prenatal development, capabilities nineteenth-century theorists would have attributed to “nature” rather than “nurture” (Moore, Reference Moore2001, Reference Moore and Kampourakis2013). But this dichotomous way of thinking about phenotype origins has been rendered obsolete by the work of developmental psychobiologists (Gottlieb, Reference Gottlieb2007; Michel & Moore, Reference Michel and Moore1995) and molecular (Lewontin, Reference Lewontin2000; Strohman, Reference Strohman2003), physiological (Noble & Noble, Reference Noble and Noble2023), and developmental (Gilbert & Epel, Reference Gilbert and Epel2015) biologists. All have conclusively established that phenotypes are emergent products of a probabilistic, multifactorial, and context-dependent developmental process that depends on the interaction and coaction of genetic and nongenetic factors. (Nongenetic factors reside in multiple places: In the environment outside the body and both inside and outside of cells but still inside the body.) The consensus of developmental systems theorists is that cognitive systems emerge from complex, dynamic interactions between – and coactions of – these genetic and nongenetic factors, where emergence is likely to be characterized by reorganization of component systems during development. Clearly, this developmental systems perspective differs radically from the more predeterministic view held by nativists like Spelke, a view that overlooks the need for investigations of the mechanisms underlying the developmental emergence of early-appearing cognitive skills.
The finding that some competencies are present at birth carries great significance for Spelke because it suggests to her that these competencies are not learned. But learning is only one component of experience, and starting with developmental processes that commence at conception, any experiential factor can potentially have profound effects on the developmental emergence of a cognitive/behavioral skill (Gottlieb, Reference Gottlieb1991; Lehrman, Reference Lehrman1953). Indeed, several experience-dependent cognitive competencies have been detected in fetuses, as Spelke discusses in chapter 9; such findings are consistent with the understanding that development is both a continuous process beginning at conception and one that normally entails reorganization.
Furthermore, some species-typical competencies are not present at birth but are no less foundational for normal functioning. Indeed, the timing of appearance of a cognitive competence need not bear any relation to how foundational it is. Therefore, developmental scientists' principal aim should not be to identify cognitive functions present at birth in order to declare them “core” and fundamental to all that emerge later. Rather, once particular functions are discovered, our job is to explicate the processes underlying their emergence in development regardless of when they emerge.
One entrenched, old idea is that there are two different processes responsible for phenotypes, one that relies on experiences and one that yields experience-independent “evolved behaviors.” This idea is known as the “phylogeny fallacy” (Lickliter & Berry, Reference Lickliter and Berry1990) simply because phenotype emergence in each generation is the product of development, and all development involves both genetic and experiential factors. To be sure, Spelke acknowledges that so-called “innate” cognitive processes must develop prenatally and that the experiences that give rise to them might involve activity generated in “subcortical or older cortical regions…that propagates to the plastic [neocortex]” (p. 195). But what brings about such prenatal activity and why should the resultant cognitive processes be considered any more “core” than earlier- or later-emerging processes? After all, the development of every higher cognitive process reflects neocortical cells' experiences with incoming stimulation. Consequently, a developmental analysis must a priori consider all stimulation as potentially crucial unless its role has been empirically ruled out. Such analyses do not need to distinguish between stimulation that arises outside of the neocortex but still within the brain (e.g., in a subcortical region) and stimulation that arises outside of the brain entirely (e.g., circulating hormones in a fetus's body, auditory stimuli that flood a fetus's brain with neurotransmitters, or a newborn's first whiff of its mother). Given this, Spelke's categorical distinction between different kinds of stimuli can be considered arbitrary and offers few insights into the emergence of cognitive skills.
At the heart of Spelke's theoretical argument are these statements: “the core systems … have been shaped by hundreds of millions of years of cognitive evolution. Some core systems are shared by animals as remotely related to us as fish, and aspects of these systems are shared by flies and worms” (p. xx). For example, “the place system is innate. Cognitive and brain scientists have studied the mechanisms and processes by which place representations arise in infant minds, through research on animals who have been reared under systematically controlled conditions” (p. 139). These statements reflect at least three significant misunderstandings.
First, every cognitive system in extant organisms reflects millions of years of evolution and relies on subsystems that have survived natural selection (e.g., functioning neurons, sodium–potassium pumps, etc.). Phenotypes with long evolutionary histories should not be considered any more “core” – in the sense of “atomic” – than more recently evolved phenotypes that contribute to organisms' survival. All cognitive systems are built from smaller components that have long evolutionary histories.
Second, Spelke writes that core systems “have been shaped by…evolution,” a claim that reflects a misunderstanding common among psychologists. This idea conflates the dynamics of populations, which can be influenced by natural selection, with the dynamics of individuals, which cannot (Witherington, Lickliter, & Moore, Reference Witherington, Lickliter and Moore2023). Natural selection does not have the creative power to “shape” individuals' phenotypes (Sober, Reference Sober1984). Although natural selection can, over time, affect the distribution of phenotypes in a population, individuals' phenotypes always reflect the developmental process that is of paramount concern to developmental scientists (Moore, Reference Moore2008).
Finally, investigating how representations arise in human infant minds by doing experiments on nonhuman animals is fraught, because homology is not identity. Bird wings are homologous with human arms, but the bones in bird wings are different in key respects from the bones in our arms (Dumont, Reference Dumont2010). Thus, Spelke is on shaky ground when she claims that the human navigational system is “core” because other animals possess such a system at birth.
Spelke is correct that her “core” systems support children's later learning, and her claim that it is advantageous to develop the capacity for abstraction early in life is insightful; abstraction need not await arrival at some developmental pinnacle. Still, there is no reason to argue that experiences are unnecessary as we develop the ability to abstract information. Although Spelke argues that infants “are predisposed to learn to … categorize objects by using shape descriptions that capture the characteristic forms of plants and animals” (p. 202), there is strong experimental evidence that some of these forms are experience-dependent. For example, the basic categories of faces (e.g., same-race vs. other-race) and language (native vs. nonnative) are acquired through perceptual narrowing during infancy (Lewkowicz & Ghazanfar, Reference Lewkowicz and Ghazanfar2009). Initially, these forms are so broadly specified that newborns do not distinguish human from nonhuman faces (Di Giorgio, Leo, Pascalis, & Simion, Reference Di Giorgio, Leo, Pascalis and Simion2012) and integrate nonhuman faces with temporally synchronized tones (Lewkowicz, Leo, & Simion, Reference Lewkowicz, Leo and Simion2010). These findings challenge Spelke's view that infants are predisposed to recognize “people as social beings” (p. 301). Clearly, that level of cognitive specificity emerges after months of everyday postnatal experience.
Is it worth knowing that a particular ability is functional at birth? Absolutely. But that does not tell us (1) how the ability develops, (2) how important it is, (3) that it develops independently of contextual factors, or (4) that it is any more “core” than later-appearing abilities. Abilities that are important to normal human functioning appear at various times in development: The ability to hear and to suck is present in utero, the ability to navigate is present at birth, and the ability to walk is present only after about a year of postnatal development. There is little about appearing at birth that makes a characteristic special. Furthermore, it is not yet possible to do fair tests of human newborns' competencies in most of Spelke's “core” domains. Consequently, Spelke draws many of her conclusions from studies of older infants, toddlers, or young children. The problem is that all these studies reflect postnatal experiences and therefore do not convincingly demonstrate that these competencies are “innate” in humans.
When Spelke's arguments invoke the commonalities we share with fish, she claims that it is “the same cognitive system, inherited from a distant common ancestor, [that] underlies their performance” (p. 129). However, vertebrates and other complex organisms inherit only developmental resources (including genes used in protein production), resources that we use to build our phenotypes anew in each generation via probabilistic developmental processes; we cannot inherit full blown phenotypes (Lewkowicz, Reference Lewkowicz2011; Moore & Lickliter, Reference Moore and Lickliter2023). In discussing the navigational abilities of fish, Spelke supports her arguments with the results of experiments that deny fish certain experiences. These are the same sorts of “isolation” experiments that Lehrman (Reference Lehrman1953) warned can only reveal if withheld environmental factors are probably not directly involved in the development of a behavior; isolation experiments do not license the conclusion that the behavior under investigation is innate. Spelke's conclusion that such experiments reveal innateness only stymies true developmental analysis.
Spelke believes “the core systems provide the foundations for the abstract concepts at the center of all our explicit knowledge” (p. xix). Perhaps. Nevertheless, this does not free developmental scientists from the responsibility of investigating the developmental emergence of such systems. Of course, it is not Spelke's personal responsibility to do so. Nonetheless, it is one thing to choose a starting point and try to explain what develops from that point, but it is quite another to imply that the developmental processes responsible for bringing about that starting point do not warrant study. Spelke effectively does the latter when she labels these competencies “core.”
The renowned Swiss professor of anatomy and physiology Wilhelm His wrote in 1888:
The single word “heredity” cannot dispense science from the duty of making every possible inquiry into the mechanism of organic growth… . To think that heredity will build organic beings without mechanical means is a piece of unscientific mysticism… . A direct explanation … [of the emergence of phenotypes] can only come from the immediate study of the different phases of individual development. (p. 295)
His highlighted the critical role of developmental analysis, regardless of whether the phenotype has a long or short evolutionary history and regardless of when in ontogeny the phenotype appears. Accordingly, we can reasonably ask nativists why we should explore how children accomplish tasks that confront them at 1 year of age (as Spelke will do in How children learn), but merely accept earlier-appearing competencies as somehow “core” and consequently not requiring developmental analysis. Scientists interested in development should resist the notion of “core” competencies because it short-circuits developmental analysis, leaving us ignorant of the factors that lead to the emergence of these competencies in the first place. Because knowledge of these factors could very well be important for understanding why adult-like competence in domains such as social communication fail to develop (as in autism), understanding their developmental origins should be a central goal of our discipline.
Financial support
This research received no specific grant from any funding agency, commercial, or not-for-profit sectors.
Competing interest
None.