RECOMMENDED READ: "Reactive inhibitory control precedes overt stuttering events" (2024)

RECOMMENDED READ: "Reactive inhibitory control precedes overt stuttering events" (2024) Research link: [https://direct.mit.edu/nol/article/5/2/432/119560](https://direct.mit.edu/nol/article/5/2/432/119560) Research points to neurofunctional differences underlying fluent speech between stutterers and non-stutterers. Considerably less work has focused on processes that underlie stuttered vs. fluent speech. Additionally, most of this research has focused on speech motor processes despite contributions from cognitive processes prior to the onset of stuttered speech. We used MEG to test the hypothesis that reactive inhibitory control is triggered prior to stuttered speech. Twenty-nine stutterers completed a delayed-response task that featured a cue (prior to a go cue) signaling the imminent requirement to produce a word that was either stuttered or fluent. Consistent with our hypothesis, we observed increased beta power likely emanating from the right pre-supplementary motor area (R-preSMA)-an area implicated in reactive inhibitory control-in response to the cue preceding stuttered vs. fluent productions. Beta power differences between stuttered and fluent trials correlated with stuttering severity and participants' percentage of trials stuttered increased exponentially with beta power in the R-preSMA. Trial-by-trial beta power modulations in the R-preSMA following the cue predicted whether a trial would be stuttered or fluent. Stuttered trials were also associated with delayed speech onset suggesting an overall slowing or freezing of the speech motor system that may be a consequence of inhibitory control. Post-hoc analyses revealed that independently generated anticipated words were associated with greater beta power and more stuttering than researcher-assisted anticipated words, pointing to a relationship between self-perceived likelihood of stuttering (i.e., anticipation) and inhibitory control. This work offers a neurocognitive account of stuttering by characterizing cognitive processes that precede overt stuttering events. **Conclusion** This study is the largest and most balanced neurofunctional investigation of stuttered versus fluent speech to date. We found elevated beta activity likely originating in the R-preSMA preceding stuttered versus fluent speech which was related to stuttering severity and proportion of trials stuttered. Stuttered trials were also associated with delayed speech onset. These neural and behavioral results suggest a reactive inhibitory control response to cues that signal to stutterers that they will soon have to produce a word that will likely be stuttered, which is consistent with previous accounts of freezing in stuttering. A post-hoc analysis also pointed to a relationship between stuttering anticipation and this reactive inhibitory control response such that words that were associated with greater anticipation (i.e., independently generated vs. researcher-assisted anticipated words) were related to greater beta power and more stuttering. This exploratory finding suggests that **anticipation precedes reactive inhibitory control** such that the inhibitory response is triggered because the speaker can predict and is averse to upcoming overt stuttering. While elevated inhibitory control is likely not the cause of stuttering events, it may contribute to stuttering events by interfering with speech motor control (e.g., thinking consciously about an automatic process) and ultimately make it more difficult for stutterers to move forward during speech. Broadly, the current results support a neurocognitive perspective of stuttering events by characterizing cognitive processes that precede overtly stuttered speech. Future work should continue to investigate the role of inhibitory control in stuttering events as well as explore the possibility of targeting the neural regions underlying hyperactive inhibitory control to facilitate speech production in stutterers. **Anticipation triggers inhibition** It is likely that an inhibitory response is triggered prior to overtly stuttered speech due to the negative consequences that stutterers associate with overt stuttering ([Bloodstein, 1972](javascript:;); [Garcia-Barrera & Davidow, 2015](javascript:;); [Jackson et al., 2015](javascript:;)) and their ability to predict it ([Knott et al., 1937](javascript:;); [Milisen, 1938](javascript:;); [Van Riper, 1936](javascript:;)). Neural evidence for such a response comes from a recent investigation by [Jackson et al. (2022)](javascript:;), which revealed increased right dorsolateral prefrontal cortex (R-DLPFC) activity seconds prior to producing anticipated versus unanticipated words ([Jackson et al., 2022](javascript:;)). This activity is consistent with a proactive inhibitory control response to stuttering anticipation. *Proactive inhibitory control* is a forward-looking process aimed at preventing or delaying undesired actions and associated with heightened R-DLPFC activity ([Aron, 2011](javascript:;); [Jahanshahi et al., 2015](javascript:;)). Overt stuttering is an undesired action because of its associated negative consequences. Therefore, individuals who stutter employ proactive inhibitory control to cope with impending stuttering and will often implement learned strategies, such as modifying the speech plan, to circumvent the anticipated word (e.g., by stalling or switching a word; [Jackson et al., 2015](javascript:;); [Jackson et al., 2018](javascript:;); [Jackson et al., 2019](javascript:;)). This is consistent with the finding of delayed speech initiation times for anticipated words in [Jackson et al. (2022)](javascript:;). It has also been proposed that reactive inhibitory control plays a role in stuttering ([Hannah & Aron, 2021](javascript:;); [Neef et al., 2016](javascript:;); [Neef et al., 2018](javascript:;)). In contrast to proactive inhibitory control, *reactive inhibitory control* refers to an automatic and rapid response triggered by an external cue to halt a planned action. Reactive inhibitory control is typically tested using stop signal tasks during which participants’ neural responses to external signals to stop an initiated action are measured. The most consistent neural marker of reactive inhibition is increased beta power in regions of the action-stopping network, including the right pre-supplementary motor area (R-preSMA), right inferior frontal gyrus (R-IFG), and subthalamic nucleus ([Aron, 2011](javascript:;); [Wessel & Aron, 2017](javascript:;)). Unlike proactive inhibitory control which may arise when a stutterer thinks about having to produce an anticipated word, reactive inhibitory control may be more immediate and triggered by an external cue that signals that overt stuttering is imminent. Consider a situation in which a stutterer is going to introduce themself to a new person. As they approach the new person, the stutterer knows they will have to say their name, which is an anticipated word for many stutterers. Leading up to production of the word, the stutterer may delay speech initiation while thinking of strategies to circumvent overt stuttering, such as using a starter utterance (e.g., “My name is \_\_\_” or, “um”). These behaviors are likely reflective of proactive inhibitory control given their prospective nature. At the same time, the stutterer may exhibit reactive inhibitory control, for example, when the new person extends their hand and introduces themself because, at this moment, the stutterer knows that they will have to produce their name imminently. In this example, the gesture of the new person (e.g., the extended hand) serves as an external cue that the stutterer will soon have to produce their name, an action to which the stutterer is averse. This cue may trigger a “freezing” response to the threat of producing the anticipated word ([Alm, 2004](javascript:;)). Due to its immediacy and automaticity, the cue indicating that the speaker will have to produce their name imminently likely triggers reactive rather than proactive inhibitory control. In the current study, we investigated reactive inhibitory control responses prior to overt stuttering events using MEG. MEG is an ideal functional neuroimaging technique to study modulations in beta oscillatory activity, which are a hallmark characteristic of reactive inhibitory control ([Alegre et al., 2004](javascript:;); [Aron, 2011](javascript:;); [Swann et al., 2009](javascript:;)). We employed our recently introduced method to elicit a semi-balanced amount of stuttered and fluent speech across participants ([Goldfarb et al., 2023](javascript:;); [Jackson et al., 2020](javascript:;)). **Discussion:** The primary finding of this study is that, compared to fluent speech, stuttered speech is preceded by increased beta power in the R-preSMA following a cue indicating that speech is imminent. Given its transient nature and timing relative to the cue, this neural response is consistent with the hypothesized reactive inhibitory control or a freezing-like response ([Alm, 2004](javascript:;)). Its cortical origin is also consistent with this interpretation, limitations in MEG source localization notwithstanding (see [Limitations](https://direct.mit.edu/nol/article/5/2/432/119560#sec23), below). Indeed, reactive inhibitory control is associated with increased beta power in the R-preSMA in response to stop signals ([Aron, 2011](javascript:;); [Nachev et al., 2007](javascript:;); [Wessel & Aron, 2017](javascript:;)). During classic stop-signal tasks, a go signal is displayed in each trial, prompting participants to perform a specific action (e.g., press a button). For a small proportion of trials, a stop signal is introduced shortly after the go signal, requiring the individual to stop their automatized response. Elevated activation in the R-preSMA is understood as a reaction to the stop signal because of its automatic and fast nature, typically occurring within 120 ms after the stop signal ([Hannah & Aron, 2021](javascript:;); [Hannah et al., 2020](javascript:;); [Jana et al., 2020](javascript:;)). A similar neural response is elicited in go/no-go tasks in which a habitual response to go signals needs to be inhibited in a small proportion of (no-go) trials ([Aron, 2011](javascript:;)). In the current study we reasoned that a cue signaling the requirement to imminently produce a word likely to be stuttered would act as an implicit stop or no-go signal, and thus result in a similar reactive and automatic inhibitory control response because the stutterer does not want to stutter overtly. Our data are consistent with this hypothesis. While we observed greater activity in the R-preSMA, we did not find elevated activity in the R-IFG. Elevated activity in the R-IFG has been associated with reactive inhibitory control ([Aron et al., 2007](javascript:;); [Aron & Poldrack, 2006](javascript:;); [Chikazoe et al., 2009](javascript:;)) and is also a common finding in studies comparing the fluent speech of stutterers and non-stutterers (e.g., [Neef et al., 2016](javascript:;); [Neumann et al., 2003](javascript:;); [Watkins et al., 2008](javascript:;)). It is unclear why the current result was restricted to the R-preSMA; however, it may have to do with the stuttering state. While the R-IFG is considered to be a neural signature of stuttering ([Belyk et al., 2017](javascript:;); [Budde et al., 2014](javascript:;)), this is primarily the case when stutterers are speaking fluently. In addition, previous studies assessed activation during or time-locked to speech execution compared to pre-speech and time-locked to a cue as in the current study. It is likely, for example, that both the aversion to stuttered speech (prior to execution) and consequent motor inhibition are absent prior to fluent speech. It may be that previous studies showing elevated activation in the R-IFG point to compensatory processing that facilitates fluent speech in stutterers which is not present when speech is stuttered. In addition, the preSMA in conjunction with the subthalmic nucleus (STN; [Cavanagh et al., 2011](javascript:;); [Herz et al., 2018](javascript:;); [Wiecki & Frank, 2013](javascript:;)), and not the R-IFG, has been implicated in other movement-related disorders. For example, reduced preSMA–STN connectivity is associated with freezing of gait in individuals with Parkinson’s disease when they walk through a doorway, which was also associated with longer footstep latency. Given that the relationship between the R-IFG and preSMA in action-stopping remains unclear, future studies can probe this relationship, for example, by focusing on disorders like stuttering to help determine the respective roles of the R-preSMA and R-IFG. Reactive inhibitory control differs from proactive inhibitory control which is forward-looking and more deliberate—for example, **the speaker has more time to reflect on the sense that upcoming speech will be stuttered**. Neurally, these types of inhibition are thought to be mediated through distinct pathways (hyperdirect and indirect BGTC pathways, respectively) and, consequently, at different rates (faster and slower, respectively). Proactive inhibitory control could occur seconds or longer before an expected action, whereas reactive inhibitory control would **immediately follow some kind of internal or external cue associated with halting the currently planned action**. Still, it is likely that proactive and reactive inhibitory control are related and overlapping processes ([Aron, 2011](javascript:;); [Liebrand et al., 2017](javascript:;)). In the context of stuttering, [Jackson et al. (2022)](javascript:;) found evidence for proactive inhibitory control preceding the production of words likely to be stuttered. In that study, there was a 5 s delay between the time when participants saw the word that they would have to produce and the go signal ([Jackson et al., 2022](javascript:;)). Although the current experiment was designed to reveal a reactive inhibitory response, it is possible that proactive control was similarly initiated soon after the words (which were subsequently stuttered) were presented, and sustained until the word was produced. This is in line with the activity found in the R-DLPFC in the current study, albeit occurring after the inhibitory control response to the cue. Alternatively, R-DLPFC activity could reflect error monitoring responses to a recognition that something has gone awry during the speech planning process ([Jackson et al., 2022](javascript:;)). Generally, our findings are consistent with proposals that hyperactive inhibitory control in the right hemisphere BGTC loop is involved in stuttering ([Hannah & Aron, 2021](javascript:;); [Korzeczek et al., 2022](javascript:;); [Neef et al., 2016](javascript:;); [Neef et al., 2018](javascript:;)). These largely speech motor-based accounts propose that inhibitory control could cause the interruptions in speech that characterize overt stuttering (i.e., blocks, prolongations, repetitions) by suppressing the initiation or sequencing of speech motor movements ([Neef et al., 2018](javascript:;)). This could occur by interfering with the forward modeling of sensory predictions required for online speech motor control (in line with [Max & Daliri, 2019](javascript:;)). Deficient forward modeling would contribute to inaccurate auditory feedback, consistent with Chang and Guenther’s account [(2020)](javascript:;). Our results offer some support for these ideas, showing divergent activity preceding stuttered versus fluent productions in the left posterior temporal cortex and SMA, that is, regions associated with the processing of auditory feedback ([Guenther, 2016](javascript:;)). However, proponents of the hyperactive inhibitory control account do not specify why reactive inhibitory control would arise in the first place, suggesting that it may occur due to random neural fluctuations in the BGTC loop. While this is plausible, it is difficult to find a reason for why this would be the case. Our account is that during development, stutterers learn to anticipate stuttering events by associating certain words or sounds with instances of stuttering and their negative consequences ([Arenas, 2017](javascript:;); [Garcia-Barrera & Davidow, 2015](javascript:;)). We argue that when stuttering on a particular word or sound is anticipated, its production is inhibited because the speaker does not want to stutter in order to avoid these negative consequences. This is consistent with our findings that (a) speech onset time for stuttered words is delayed and that (b) independently generated anticipated words are related to both more stuttering and greater beta power than researcher-assisted words because independently generated words are more likely to trigger these negative associations. Once these associations are in place, inhibitory control may act proactively, for example, when the speaker knows they will have to produce a word likely to be stuttered (as in [Jackson et al., 2022](javascript:;)), or reactively, triggered by a cue (implicit or explicit) signaling the requirement to imminently produce the anticipated word. Importantly, we believe that stuttering can occur in the absence of elevated inhibitory control, particularly during spontaneous speech when either there is not enough time to retrieve these associations (i.e., to anticipate stuttering) or the speaker’s attention is not directed toward the speech production process. Critically, the paradigm in the current study created a setting in which it was likely that if a word was stuttered, it was also anticipated. As a result, elevated inhibitory control was also likely. The current findings, as well as other recent work ([Jackson et al., 2022](javascript:;)), have important implications for neuromodulation as a possible complement to existing behavioral therapy. Transcranial direct current stimulation (tDCS) is starting to be applied in stuttering research ([Chesters et al., 2018](javascript:;); [Garnett et al., 2019](javascript:;); [Yada et al., 2019](javascript:;)), albeit with modest results. For example, [Garnett et al. (2019)](javascript:;) tested the impact of anodal tDCS on overt severity in 14 adult stutterers, and while they did not find significant effects on overt stuttering severity, they found that the atypically strong association between overt severity and right thalamocortical activity was attenuated after tDCS, especially in severe stutterers. It may be that the modest effects reported to date are due to an exclusive focus on the speech network. **Future neuromodulation studies can target, for example, proactive (R-DLPFC) and reactive inhibition (R-preSMA) to test whether forward-moving speech is facilitated by reducing the putative interference from hyperactive inhibitory control on speech production.**