When Sommer et al. (2002) published their findings about white matter deficits in the brain of adult stutterers, they hoped they had discovered the physical cause of stuttering: a disturbed signal transmission between speech-relevant brain areas, namely the temporo-parietal region, on the one side, and the lateral pre-motor and motor areas of speech control, on the other side– roughly said, between Wernicke’s and Broca’s areas (see also Büchel & Sommer, 2004). What they precisely had found was a reduced fractional anisotropy in the left superior longitudinal fasciculus (SLF), a white matter tract connecting the brain regions mentioned. Similar results were obtained in further studies in adults who stutter (Cai et al., 2014b: Connally et al., 2014; Cykowski et al., 2010; Watkins et al., 2008).as well ss in stuttering children (Chang et al., 2008; Chang, Zhu, Choo, and Angstadt, 2015; Chow and Chang, 2017)
Figure 13: Charting of the superior longitudinal fasciculus (SLF) on the left brain hemisphere, connecting the posterior superior temporal cortex, the supramarginal gyrus (SMG) on the inferior parietal cortex, and the posterior part of Broca’s area (pars opercularis, BA44) (read more). MC = motor cortex, SC = somatosensory cortex (the parts responsible for speech movements among others).
A reduced fractional anisotropy in white matter is mainly interpreted as a reduced density or a reduced structural integrity of the nerve fibers, especially as a lower degree of myelination. Indeed, fractional anisotropy could also be reduced because of crossing fibers, a lower number of fibers, or an uneven structure of the bundles, but I will neglect these possibilities in the following discussion mainly because I do not believe that such relatively invariable structural conditions come into question as causal for a variable disorder like stuttering.
The reduced fractional anisotropy found in the SLF in stutterers suggests a lower degree of myelination, or, in other words, of maturation of the nerve fibers. Nerve fibers are not simply wires between neurons, but axons, in other words, they are the neurons themselves, since the axon is the largest part of a neuron in the brain, thus the activity of a neuron is the activity of its axon. This is not unimportant for understanding the mechanism of myelination.
Myelination proceeds in the brain in the following way: Oligodendrocytes, a sort of glia cells, wind themselves in layers around the axon. The myelin sheath increases the speed at which impulses propagate along the axon: The saltatory conduction of impulses in a myelinated axon is up to 16-times faster than the continuous conduction in an unmyelinated axon.
Myelination occurs during different times and with different speed in different parts of the brain. In the speech-related network, it proceeds mainly during childhood and adolescence, that is, during the time of language acquisition and development; so Brauer (2009)) found a significantly reduced fractional anisotropy in the SLF in seven-year-old normal fluent children, compared to adults. Therefore, the reduced fractional anisotropy in some brain regions found in stutterers suggests, that myelination is either delayed or impaired there. However, the findings do not tell us whether they are causal for stuttering or only a consequence or concomitant of the disorder – that is, whether they are the chicken or the egg. To address this question, it is helpful to know how myelination is controlled, or what it is influenced by.
Bengtsson et al. (2005) investigated the effect of piano practicing in childhood, adolescence and adulthood on white matter. They found a positive correlation between practicing and fractional anisotropy in different brain regions involved in piano playing. Scholz et al. (2009) conducted a longitudinal study with individuals who learned a novel visuo-motor skill – juggling. The training group and an untrained control group were scanned before and after a six-week training period. The authors detected a significant increase of fractional anisotropy in white matter underlying cortex regions involved in the control of juggling in the trained group.
A similar study was carried out by Keller and Just (2009): They determined whether 100 hours of intensive remedial instruction affected the white matter of 8- to 10-year-old poor readers. The intensive practice resulted in a significantly increased fractional anisotropy in a brain region in which fractional anisotropy was lower on average in the poor readers, compared with good readers. The increase of fractional anisotropy was correlated with a decrease in radial diffusivity (but not with a change in axial diffusivity), suggesting enhanced myelination – and it was correlated with an improvement in reading. Further, Takeuchi et al. (2010) showed that the amount of working memory training correlated with increased fractional anisotropy in white matter regions that are thought to be critical in working memory. In this study, young adults practiced memorizing the spatial and temporal order of visually presented stimuli, 25 minutes a day over two months.
These results suggest a relationship between activity over time and myelination: The more frequently axons (that is, neurons) are activated, the better they become myelinated – just like a muscle that gradually becomes stronger by training. This assumption is confirmed by molecular biology: Wake, Lee, and Fields (2011), with axons of mice in culture, observed a mechanism able to promote the myelination of electrically active axons: The electrical activity of the axon provides an instructive signal to the oligodendrocytes for producing the myelin basic protein. That means: The more frequently an axon is active, that is, the more frequently the neuron ‘fires’, the better it becomes myelinated (see also Fields, 2010, Zatorre, Fields, & Johansen-Berg, 2012) (read more).
However, what is the sense and purpose of this connection between axonal activity and myelination? As mentioned above, myelination accelerates the conduction of impulses along the axon. The effect is not so much that the organism more quickly reacts, but that the fastest neuronal network can suppress competitive networks and, by this, determine the organism’s spontaneous behavior. In this way, a behavior frequently repeated is automatized with time, since the neuronal network controlling this behavior is more frequently active and becomes better myelinated. That makes sense from the viewpoint of evolution because, in wilderness, a behavior frequently repeated can hardly be a wrong behavior, i.e., a behavior not conducive to survival. Therefore, myelination modulates neural development according to an organism’s environmental experience – and it supports a non-cognitive learning, a learning by repetitive doing, that is, the stabilization and automatization of abilities, the formation of useful routines (read more).
Learning by repetition, automatization, and habituation play a central role also in speaking. In this way, a child learns to articulate words, to conjugate them, and to put them together to clauses in the correct order – without any explicit knowledge of grammar or syntax. The child simply imitates his/her parents’ or siblings’ behaviors and habituates them. As was already emphasized in Chapter 1, spontaneous fluent speech is much more a matter of behavioral routines than of conscious planning. It is, therefore, plausible to assume that the gradual myelination of the speech-related neuronal network accompanies the acquisition of one’s native language.
Let us, now, return to the issue of why some parts of the left SLF in stutterers could be less myelinated: If nerve fibers frequently active are preferentially myelinated, then ,in turn, fibers that have seldom been active over time will be less myelinated. Therefore, the deficits found in the left SLF may result from less activity of these fibers, that is, from less activity of the posterior superior temporal cortex (Wernicke’s area) and/or the inferior frontal cortex (Broca’s area), since the affected SLF fibers might mainly be the axons of neurons localized in these cortical areas. And just this – an under-activation mainly in the posterior superior temporal cortex, but also in the inferior frontal cortex in stutterers during speech – was found in several brain imaging studies (e.g., Braun et al., 1997; Brown et al. (2005); Budde, Barron, and Fox (2014); Fox et al., 1996; Ingham et al., 2003). Therefore, the myelination of the fibers connecting Wernicke’s and Broca’s area may be delayed in stutterers:
After the present theory, stutterers direct too little attention to the auditory channel during speech, hence secondary auditory areas are under-activated, with the consequence that also the fibers connecting this brain region with the region of speech control in the inferior frontal cortex are less active, compared to normal fluent individuals. Long-standing under-activation may cause a delay in myelination.
That means, stuttering as well as the structural deficits in the left SLF may be consequences of the misallocation of attention during speech. Then, the low myelination is neither a cause nor a consequence, but a predictor of a predisposition for stuttering. However, some differences in myelination in other brain regions between stutterers and nonstutterers (see, e.g., Cai et al., 2014b: Connally et al., 2014) may be consequences of longstanding stuttering, namely the result of habituated secondary behaviors.
As mentioned above, myelination – also referred to as the maturation of the fibers – proceeds gradually over time, with the effect that impulse conduction becomes faster. However, also a non- or less myelinated axon is able to work – the time difference is only a few milliseconds because of the short distances in the brain. A nerve fiber poorly myelinated because of less activity is a healthy fiber – similar to a muscle that is weak because it has not been trained. That means, the fiber connection between speech perception (Wernicke’s area) and speech control (Broca’s area) is unimpaired and able to work. We can observe this in conditions in which stutterers are required to involve auditory information into speech control: in speaking in sync with the beat of a metronome, in chorus reading, and in shadowing (see Section 3.1). Not only that stutterers, in a rule, are able to accomplish these tasks – even their stuttering mostly disappears completely or is markedly reduced in these conditions.
In brain research, fractional anisotropy is a measure describing the degree of anisotropy in the diffusion of water molecules. A value of zero means that diffusion is isotropic, i.e. it is equal in all directions. In and around nerve fibers, diffusion is anisotropic, namely lower orthogonal to the direction of the fibers and/or higher in fiber direction. Tendentially, nisotropy is the higher the better the fibers are coated with myelin – even if also some other factors influence fractional anisotropy, e.g., crossing fibers. Particularly radial diffusivity, i.e., perpendicular to fiber direction, is clearly correlated to myelination: the lower the radial diffusivity, the thicker the myelin. The relation of axial diffusivity, i.e., in fiber direction, to myelination, or to the maturation of the fibers in general, is still a matter of debate (see, e.g., Alexander et al., 2007; Bartzokis et al., 2012; Mädler et al., 2008; Uda et al., 2015; Wei et al., 2013).
The precise anatomy of the dorsal fiber tracts connecting temporal/parietal cortex and frontal cortex are still a matter of research and debate (compare Catani, Jones, & ffytche, 2005 with Makris et al., 2005). Figure 13 very roughly symbolizes SLF III and/or arcuate fasciculus, the tracts in which a lower fractional anisotropy in stutterers mainly was found. SLF III runs horizontally up to the angular gyrus, but this region seems to be more involved in the association of language with visual processing during reading and writing. For more details see Fig. 5,2 in Brauer (2009), Fig. 5 in Catani & Mesulam (2008), Fig. 8 in Frey et al. (2008), Fig. 1 in Friederici (2012), Fig. 6 in Kelly et al. (2010), Fig. 3 and 4 in Makris et al. (2005).
“Myelin formation requires cell recognition to myelinate the appropriate axon, the formation of adhesive contacts, elaboration of vast areas of cell membrane to form myelin sheets, wrapping multiple layers of membrane around axons, and the removal of cytoplasm from between the wraps of myelin to form compact stacks of lipid membrane, all of which might be influenced by signaling from electrical activity in axons. […] This would preferentially myelinate axons that are electrically active and increase the speed of conduction through these functionally active circuits. This process could therefore underlie some of the changes in white matter seen in MRI studies.” (Zatorre, Fields, & Johansen-Berg, 2012, p. 8 in the PDF)
A myelin sheath around an axon makes impulse conduction faster, but it can hardly be the sense of the progressing myelination from birth to adulthood to make the brain faster as a whole – if it was so, all axons could be readily myelinated by birth, as it is the case with some tracts that are, so to say, part of the operating system of the brain. Therefore, the assumption that myelination serves for the adaptation of an organism’s behavior to a specific environment is very plausible.
Above I wrote that, from the viewpoint of evolution, it makes sense that a frequently repeated behavior becomes automatized because, in wilderness, a behavior frequently repeated can hardly be a wrong behavior. Here, an additional remark is necessary: Whereof does a wildlife animal learn what kind of behavior is right and should be repeated, and what is wrong and should not be repeated? The only way of learning is by reward or punishment immediately following a behavior: either feed, sex, safety – or hunger, thirst, frustration, fear, pain. Thus it is an animal’s environmental experience that forms behavior, and the frequently repeated behavior forms the brain. In this way, successful behaviors become stabilized and automatized.
Of course, the same mechanism works in humans, which has different consequences. In the acquisition of new abilities, e.g., in crafts, arts, and sports, but also in everyday behaviors, we learn by practicing and by the immediate experience of whether it goes well or awry, easily or heavily, whether a way of doing should be repeated or not. As the results of Scholz et al. (2009) and Keller and Just (2009) suggest, the neuronal networks that control frequently repeated practices were better myelinated and, in this way, successful behaviors were stabilized and automatized. They become behavioral habits.
However, the simple biological mechanism of learning from reward or punishment immediately following a behavior is not without risk to humans in the civilization: For example, the consumption of sweets, nicotine, alcohol, and drugs is immediately rewarded; the person feels better. Likewise, lying and stealing are often immediately rewarded – punishment mostly comes much later if ever. By immediate reward and the temptation of repeating in these cases, unwholesome or socially undesirable behaviors can be habituated, and the myelination of frequently activated neuronal networks may play a role in this process.