Responses to startling acoustic stimuli indicate that movement-related activation does not build up in anticipation of action

to the editor: A recent article by Marinovic and colleagues (2013) used a loud acoustic stimulus (114 dB) to probe motor preparation prior to a visual imperative stimulus (IS) in a simple reaction time (RT) task. Results revealed decreasing RT latency and increasing motor output as the probe stimulus approached the IS, leading to the conclusion that response-related activation increased abruptly in the 65 ms before movement initiation. This result contradicts research that has shown prepared movements can be triggered by a startling stimulus (Valls-Solé et al. 1999; “StartReact” effect) at a consistently short latency well in advance of the IS (e.g., Carlsen and Mackinnon 2010). We believe that methodological discrepancies between this experiment and previous ones affected the results, and thus the conclusions offered by the authors.

It is generally accepted that the StartReact effect arises due to activation in startle-related brain stem (e.g., reticular) structures interacting with voluntary response channels to involuntarily trigger the intended action (see Carlsen et al. 2012; Valls-Solé et al. 2008 for reviews). However, it is not adequate to rely on the shortened latencies alone to infer brain stem activation. Such effects as stimulus intensity (Kohfeld 1971) and intersensory facilitation (Gielen et al. 1983) can also speed responses via voluntary response channels. Thus, to confidently attribute speeded responses to brainstem activation, separate indication of this activity is required, which can be provided by the reflexive startle response. Short-latency electromyographic (EMG) activity in the sternocleidomastoid (SCM) has been shown to be a robust indicator of sufficient brain stem activation to result in a consistent StartReact effect (Carlsen et al. 2011b) that is differentiated from the effect of stimulus intensity (Carlsen et al. 2007; see also Tresch et al. 2014) or intersensory facilitation (Carlsen et al. 2011a). Although StartReact can be dissociated from the startle reflex in a few specific cases (e.g., prepulse inhibition; Maslovat et al. 2012; Valls-Solé et al. 2005), measuring reflexive startle activity ensures a reliable indicator of response triggering. For example, by manipulating stimulus intensity, Carlsen et al. (2007) showed that trials where no startle response was elicited (as indicated by a burst of EMG in SCM; SCM−) exhibited typical stimulus intensity effects (i.e., RT reduction from ∼120 ms at 83 dB to ∼100 ms at 123 dB), whereas SCM+ trials showed a significantly lower RT latency that was similar (∼80 ms) irrespective of intensity. However, the likelihood of eliciting a startle response increased with stimulus intensity, suggesting that if SCM+ and SCM− trials were grouped together, lower intensities would be “contaminated” by a relatively high percentage of nonstartle RT latencies. The analysis by Marinovic et al. (2013) did not measure reflexive SCM activation and thus may have included trials in which the acoustic stimulus did not trigger the response, producing results from which inaccurate conclusions were made.

To reexamine the findings of Marinovic et al. (2013), we replicated their experiment with the additional dependent variable of EMG activation in the SCM (Maslovat et al. 2015). Our replication analysis, which included all trials irrespective of the presence of a startle indicator, confirmed results consistent with those by Marinovic et al. However, when trials were compared on the basis of presence or absence of startle-related SCM activation (i.e., SCM+ vs. SCM−), SCM+ trials exhibited significantly shorter RT latency and increased motor output measures (Fig. 1), characteristics typically associated with the StartReact effect (Carlsen et al. 2012). Furthermore, no differences in preparatory activation were found for SCM+ trials in the final 65 ms preceding the IS. These results lead us to conclude that 1) reflexive startle activation in SCM is a necessary indicator to confirm response triggering effects; 2) in a simple RT task, movement-related activation is constant 65 ms prior to action initiation; and 3) the results of Marinovic et al. are likely confounded by trials in which the probe stimulus did not trigger the prepared response.

Fig. 1.

Mean data (SD indicated by error bars) from Maslovat et al. (2015), based on acoustic probe lead time, separated by whether sternocleidomastoid activation was present (SCM+; filled bars) or absent (SCM−; open bars). Asterisks denote SCM+ trials that were performed at significantly shorter reaction time latency (A) and significantly higher peak electromyography (EMG; B) compared with SCM− trials. Note that probe lead time did not significantly affect reaction time (P = 0.737) or peak EMG (P = 0.928) for SCM+ trials in the 65 ms prior to the imperative stimulus (time 0).

GRANTS

This work was supported by separate Natural Sciences and Engineering Research Council of Canada (NSERC) grants awarded to I. M. Franks and A. N. Carlsen.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

D.M. prepared figures; D.M., I.M.F., and A.N.C. drafted manuscript; D.M., I.M.F., and A.N.C. edited and revised manuscript; D.M., I.M.F., and A.N.C. approved final version of manuscript.