The Adaptable Mind: What Neuroplasticity and Neural Reuse Tell Us about Language and Cognition - John Zerilli 2021
The Nature of Plastic Changes in the Brain
Aspects of Neuroplasticity
2.2.1 Definition
Learning raises an interesting question for the cognitive and neural sciences. On one hand, the nervous system appears to be wired very precisely. On the other hand, mammalian and especially human behavior can be extremely flexible. If connections between the main signaling units of the nervous system are set during early development, how is it that behavior and its neural underpinnings can be flexible at all? What is the extent of neural fixity and flexibility in early development, and how is it related to the stability and dynamism exhibited under different conditions in later life (e.g., during learning or rehabilitation)? The best answer so far attempted and empirically substantiated is the plasticity hypothesis (Kandel et al. 2013, p. 37). This recognizes what for most of the twentieth century was denied: that even after a critical period in early childhood, the brain retains its plastic potential throughout life. It appears that “chemical synapses are functionally and anatomically modified through experience and learning as much as during early development” (Kandel et al. 2013, p. 37). Plasticity is an intrinsic and persistent property of the nervous system without which it would be impossible to understand normal psychological function, or indeed pathological and contrapathological responses to events throughout life (Pascual-Leone et al. 2005, p. 378). Plasticity is not to be conceived of as an occasional or exceptional state of the nervous system—it is in fact its normal and ongoing condition (Pascual-Leone et al. 2005, p. 379). What is more, similar mechanisms appear to be at work in both adult plasticity and early development, suggesting that the mechanisms of adult learning and developmental plasticity are to some considerable extent conserved (Saitoe & Tully 2001; Kolb et al. 2001, p. 224; Neville & Bavelier 2001, p. 261). This last point is crucial, since it is really only by virtue of such parallels that adult neuroplasticity can serve as a window onto early developmental processes and carry significance for traditional debates in psychology; for example, about the innateness of language. As Laurence and Margolis observe:
Widespread and significant instances of neural plasticity suggest an inherent openness to the functions that any cortical area can take on. If this is right, then the brain’s concept acquisition capacities needn’t be innately constrained toward any particular outcome. Instead cortical circuits might simply form as required to accommodate a learner’s needs given whatever contingent sensory input has been received and the wiring that has been previously established. (2015, p. 124)
Neuroplasticity has been defined as “a change (either a strengthening or weakening) in synaptic efficacy brought about through experience” (Rose & Rankin 2001, p. 176). In fact, synaptic plasticity is only one of a family of brain plasticities falling under the general banner of neuroplasticity. In its widest sense, “neuroplasticity” refers simply to “the capacity of the nervous system to modify its organization,” especially in response to experience, and includes the varied circumstances of normal development and maturation, learning in both immature and mature organisms, recovery of function after injury, and compensation following sensory deprivation (Neville & Bavelier 2001, p. 261). At the same time, neuroplasticity transverses every level of organization in the brain, synaptic events having counterparts in both higher and lower levels of organization, running all the way from genes right through to complex behavior (Shaw & McEachern 2001). These facts should not, of course, be taken to suggest that a synaptic definition of neuroplasticity is necessarily mistaken. Indeed, it is just because the synaptic level continues to provide the best understood and arguably most powerful model of neuroplasticity available—synaptic plasticity has a probable role in all of the developmental stages just described, for instance—that it has become customary to regard synaptic plasticity as broadly representative of the phenomenon. Given my concern with modules and the likely role of synaptic plasticity in the arrangement and rearrangement of cortical circuitry (Neville & Bavelier 2001, p. 261; Shaw & McEachern 2001, p. 434), there is actually good reason for framing the discussion of neuroplasticity here in terms of synaptic plasticity. Synaptic plasticity supplies a familiar and tractable neurobiological model for understanding the cases of neuroplasticity that are likely to be of direct concern to the modularity of mind; namely, cortical reorganization and memory consolidation. Still, it is important to appreciate that the term “neuroplasticity” has a significantly wider scope than the plasticity associated with merely one level of the brain’s organization; and after a brief treatment of synaptic plasticity revealing the mechanisms underlying plastic change, I must ultimately turn to consider cortical map reorganization—an instance of neuroplasticity that ought to be prioritized in any serious discussion of modularity (Rowland & Moser 2014). (As for the relationship between modules and cortical maps, see the discussion in § 4.3.)
2.2.2 Synaptic plasticity
Neurons are the basic cellular units of the nervous system—self-sufficient, specialized cells whose primary function is to receive, integrate, and transmit information throughout the body. Any neuron will receive information from potentially many thousands of other neurons through a “synapse”—a cleft between the terminals (“axons”) and receptive fibers (“dendrites”) of adjacent neurons. Neural connections may be strengthened or weakened in a variety of ways, but the most frequently cited mechanism involves adjustments to the quantity of neurotransmitter released from the presynaptic cell and/or the number of postsynaptic receptors, which determine how effectively the postsynaptic cell can respond to the quantity of neurotransmitter released presynaptically. Strengthening occurs typically by persistent stimulation of the postsynaptic cell. A neurotransmitter’s release into the synaptic cleft initiates a cascade of biochemical events that may lead to the excitation (or “potentiation”) of the postsynaptic neuron. Research has repeatedly turned up a number of neurotransmitters, neuromodulators, and ions that appear to be crucial for synaptic plasticity, including glutamate and calcium ions (Ca++). Glutamate is among the most excitatory of neurotransmitters so far discovered and works by inducing a postsynaptic calcium influx, which, through repeated stimulation, may result in an action potential. More precisely, the influx of Ca++ leads to increases in the number and efficacy of postsynaptic α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors, themselves crucial for consolidating synaptic connections by providing the primary excitatory input drive on the postsynaptic neuron.
Initially, synaptic plasticity was thought to be limited to such molecular mechanisms alone, entailing few if any changes to the shape of dendritic spines or the number of axonal branches and sprouts (i.e., neuromorphological changes leading to “synaptogenesis” and synaptic “pruning”—the establishment of new connections and elimination of existing connections), while “neurogenesis” (the generation of new neurons) was understood to be an exclusively developmental process. It is now known that beyond enhanced signaling between neurons, synaptic plasticity routinely involves changes to neuromorphology, and that neurogenesis occurs well into adult life, not just perinatally as was once thought (Rose & Rankin 2001, p. 176; Fuchs & Flügge 2014).1
Two varieties of plasticity widely considered to involve changes at the synapse are cortical map plasticity (otherwise known as representational or topographic map plasticity) and the cellular changes attendant on learning and memory consolidation (Buonomano & Merzenich 1998). “Cortical map plasticity” refers to the detailed remodeling of cortical maps in response to “behaviourally important experiences throughout life” (Buonomano & Merzenich 1998, p. 150). Evidenced across different modalities in a significant number of mammalian species, including humans, cortical map reorganization not only results from behavioral changes, the environment, and injury in later life, but is at least partly responsible for some kinds of early perceptual and motor learning (Buonomano & Merzenich 1998, p. 150). It covers language cross-lateralization (migration of function from the left to the right hemisphere) following injury or trauma early in life (and even later in life, with adequate rehabilitative training), as well as, perhaps most especially, the plasticity of sensory and motor maps in response to use or trauma. Notice that in the context of cortical map plasticity it becomes more useful to think of plasticity as the opening and closing (or broadening and narrowing) of afferent input channels. It is this plasticity that seems to have most recently captivated philosophers (see further discussion at § 8.2.2.1).
While the cellular changes involved in learning and memory consolidation are also thought to depend on synaptic plasticity, it has been far from easy obtaining empirical confirmation of this connection, or indeed of whether the same plastic mechanisms are involved in both cortical map plasticity and memory-related synaptic plasticity (Buonomano & Merzenich 1998, p. 150). Mainstream opinion in the field seems to err on the side of an affirmative connection on both counts (Buonomano & Merzenich 1998, pp. 152, 153), but Buonomano and Merzenich (1998, p. 179, cf. p. 165) cautiously conclude that, as for the connection between synaptic plasticity and cortical map plasticity, “we do not yet have a sufficient understanding of synaptic and cellular plasticity to fully account for the experimental data on cortical representational reorganization.”
The kind of memory involved is important to clarify here. There are generally two broad classes of memories distinguished by psychologists. The first is known as “explicit” or “declarative” memory, which in turn consists of both “episodic” memory for events and “semantic” memory for concepts and facts. The second class is known as “implicit” memory, which is what could be considered the memory for actions. An important class of implicit memories, relating specifically to learned skills and habits, is called “procedural” memory—on account of its role in the performance of routine procedures involving neither deliberation nor specific memories of having carried them out previously (e.g., brushing one’s teeth, tying one’s shoelaces, riding a bicycle, etc.). Procedural memory is in effect the memory for automated action cued by specific contexts and stimuli.
While models of memory-related synaptic plasticity certainly can (and perhaps do) explain the neurophysiology of explicit memory consolidation (especially semantic memory), it is implicit and procedural memory that is perhaps their more natural target (Rose & Rankin 2001, p. 176). For one thing, procedural memory is plausibly more likely to reflect the neurophysiology of learning and memory than is something like episodic memory, which is somewhat difficult to construe in terms of the establishment and maintenance of synaptic connections. In contrast, procedural memory epitomizes the rule that “practice makes perfect,” making it the more obvious candidate for explanation in terms of altered connection strengths arising from persistent, repeated cellular activation. It is also worth noting that the theory of procedural memory answers a set of concerns similar to those addressed by the theory of modularity, which in its Fodorian guise was likewise put forward to explain targeted and automated behavior involving interconnected cortical networks.
The clearest case of synaptic plasticity, and one that is likely to play some role in, or otherwise serve as a model for, memory consolidation—and possibly many other varieties of neuroplasticity—is hippocampal long-term potentiation (LTP), which, as its name suggests, is the enduring association of neurons through repeated afferent activation in the hippocampal formation. While its role in learning and memory is not conclusively established, some such role has been conjectured to exist from its resemblance to “Hebbian plasticity,” named after the Canadian psychologist D. O. Hebb. Hebb’s (1949) influential model of plasticity was advanced to explain the long-lasting changes in synaptic strength that he hypothesized to underlie learning and memory. He assumed that stable changes in synaptic efficacy could occur through interactions among neurons:
When an axon of cell A is near enough to excite a cell B and repeatedly or persistently takes part in firing it, some growth process or metabolic change takes place in one or both cells such that A’s efficacy, as one of the cells firing B, is increased. (1949, p. 62)
Hebb’s postulate asserts that simultaneous or rapidly successive pre- and postsynaptic activity results in a strengthened connection between cells (“cells that fire together wire together; cells that fire apart wire apart,” as it is often put). This requires a “coincidence detector” that records the co-concurrent or rapidly successive activity of pre- and postsynaptic neurons (Buonomano & Merzenich 1998, p. 154). In hippocampal LTP, a subtype of the glutamate receptor, so-called N-methyl-D-aspartate (NMDA) receptor, serves this coincidence-detecting function by facilitating the postsynaptic influx of Ca++ (which, if persistent, typically results in a strengthened connection via increased AMPA receptor efficacy, as we saw earlier). Since LTP appears to reflect something like Hebbian associative plasticity, many neuroscientists have not hesitated in postulating LTP as the neurochemical basis of learning and memory. It has the unique phenomenology, induction characteristics, and longevity “to place it firmly as a candidate for the storage of experiential memory” (Teyler 2001, p. 101).2
While LTP is generally regarded as crucial to memory storage, some neuroscientists are more circumspect, either denying that the evidence of LTP’s subserving learning and memory is strong enough to justify the faith placed in this mechanism (Cain 2001, p. 126), or maintaining that LTP might instead be “a generic mechanism for increasing synaptic gain throughout the brain whenever increases in synaptic strength are needed,” and therefore “a general purpose mechanism by which synapses can increase their influence . . . regardless of the kind of circuit in which they are embedded” (Teyler 2001, p. 105). An equally pessimistic estimate has it that “if LTP occurs naturally in the behaving animal, it can at best be said to underlie circuit formation, not learning or memory” (Shaw & McEachern 2001, p. 434). LTP may then, on a minimal reading, simply be a means by which neural networks are formed and maintained. But what few would deny is that LTP is an important neurophysiological substrate supporting various manifestations of neuroplasticity.3 In fact, if the connection between implicit memory and modules is rightly drawn, LTP, even in a deflationary view, could be seen as offering support to the idea that similar synaptic mechanisms are implicated in the consolidation of memory, the development of modules, and the migration of cortical maps, which, like memories and modules, are also represented in stable, if more local, networks of neurons in the brain. (Rowland & Moser [2014] present evidence that even episodic memory has modular organization, resembling the neuroanatomical and neurophysiological features of sensory and motor cortical maps, such as columnar structure and topographic arrangement. See §§ 2.4 and 4.3 for elaboration.)