Introduction {#s1} ============ The ability to rapidly adjust posture in response to small perturbations is a critical capability for postural control in the presence of threat (Marsden et al., [@B35]; Prochazka and Ellaway, [@B45]; Trewartha et al., [@B58]). Anticipatory postural adjustments (APAs) are characterized by an *initial burst* of muscle activity, followed by a *sustained* contraction (Ono et al., [@B43]). The initial burst of muscle activity helps the person to counteract any sudden perturbation by rapidly contracting leg muscles to re-position the body (Bouisset and Do, [@B6]). The sustained phase consists of an ongoing low-amplitude (\<10% of maximal voluntary contraction; MVC) and highly selective increase in muscle activity (Naito et al., [@B42]; Bouisset and Do, [@B6]; Crenshaw and Roy, [@B14]). For example, a study examining the response of calf muscles to unpredictable perturbations reported that the soleus and gastrocnemius muscles showed a sustained activity lasting up to 70% of the perturbation phase when the perturbation amplitude was ≥20% of body mass (Bouisset and Do, [@B6]). The muscles that are most commonly studied when investigating the APAs are the ankle dorsiflexors and plantarflexors, which have a higher background activation rate in standing posture than other muscles, such as the quadriceps, hamstrings and gastrocnemius muscles (Kearney et al., [@B26]; Forner-Cordero et al., [@B20]). However, muscle activity in the calf muscles can be readily detected during even small perturbations, and the sustained period can last up to \~45% of the perturbation phase (Bouisset and Do, [@B6]; Crenshaw and Roy, [@B14]). Further, while all of the calf muscles show a burst of electromyography (EMG) activity in response to sudden perturbations, the tibialis anterior (TA) and soleus muscles show more prolonged activity in a sustained manner (Carling and Kearney, [@B10]). For example, TA activity during a sudden perturbation remained elevated for a longer time period and is higher in magnitude than soleus activity (Carling and Kearney, [@B10]). These features have prompted some researchers to believe that the TA may be more involved in postural stabilization than the soleus (Carling and Kearney, [@B10]). It has been reported that TA activity is higher than soleus activity in a postural preparation state with eyes open in response to unpredictable perturbations (Mochizuki et al., [@B40]; Crenshaw and Roy, [@B14]). The APAs are often considered important for providing a compensatory action against threat. Since the APAs are highly dependent on proprioceptive input, sensory perturbations (e.g., lower limb vibration) can be used to test the response to an impending fall (Zanone and Nussbaum, [@B64]). For example, vibration of the lower leg muscles increases the muscle activity in the leg muscles during standing (Grangeon et al., [@B22]) and reduces postural steadiness (Gill et al., [@B21]). The amount of time spent in the late phase of muscle activity is proportional to the degree of perturbation, and this relationship has been used as a basis for fall risk evaluation (Kearney et al., [@B26]; Crenshaw and Roy, [@B14]; De Freitas et al., [@B15]). Additionally, a recent study reported that greater TA muscle activity can be considered to indicate greater risk for falls and future falls in older adults (Lajoie et al., [@B31]). However, the traditional definition of an APA is not sufficient enough to describe this phenomenon as it does not cover the entire response of the muscles to perturbation. In fact, the APA should be viewed as a multi-phase process that can be divided into phases based on different attributes of the activity. For example, the initial peak response can be separated into two phases: rapid onset (rapid response) and rapid offset (rapid decay; Bouisset and Do, [@B6]). This notion was further strengthened by a recent study which showed that the onset and offset phases of EMG activity were not synchronous and had non-linear, non-linear (NNL) relation (Arya and Kearney, [@B1]). Furthermore, the sustained phase of muscle activity showed a non-linear, NNL relationship with the perturbation frequency and magnitude (Arya and Kearney, [@B1]). While the current definition provides a good estimate of the magnitude of the APA in response to a perturbation, it is not sufficient to characterize the APA response during a sustained phase, where a sustained response to a perturbation might become critical for postural control. Additionally, although the APAs in the initial burst and sustained response to perturbation have been investigated separately, there is a lack of investigation of the APAs in response to perturbations which vary at different magnitudes. Further, APAs in response to a postural perturbation are often studied in healthy young adults, and there are no studies on the APAs in response to a perturbation that is relevant for people with neurological disorders or other health conditions which often have balance difficulties. Therefore, the purpose of the present study was to investigate the initial burst and sustained response to perturbation in different magnitudes in people with chronic stroke. An alternative description of the APA can be obtained from the "threshold concept" (Vereijken and Kerkhof, [@B61]), where the APA is divided into two phases: movement initiation phase (MIP) and movement adjustment phase (MAP). Specifically, the MIP corresponds to the amount of time in which the muscle is primed to react to a perturbation (Carling and Kearney, [@B10]). Similarly, the MAP corresponds to the amount of time needed for the activation of the muscles to reach a level that can counteract the perturbation (Carling and Kearney, [@B10]). The MIP-MAP relationship is also observed in single-joint, rapid unloaded joint movements and can be used to examine the APAs in response to perturbations (De Freitas et al., [@B15]; Jaspers et al., [@B25]). Specifically, it has been reported that the perturbation magnitudes can influence both phases of the MIP-MAP relationship (De Freitas et al., [@B15]). For example, a recent study reported that the perturbation magnitude strongly influences the rate of recruitment in people with stroke (Jaspers et al., [@B25]). Similarly, a study that examined the relationship between APAs and perturbation magnitudes in response to a single leg stance found that the slope of the MIP-MAP relationship became larger when the perturbation amplitude increased (De Freitas et al., [@B15]). Thus, it is important to understand the MIP-MAP relationship in the APAs in response to perturbation to fully appreciate the nature of APAs. In order to fully understand the MIP-MAP relationship in the APAs in response to a perturbation, one needs to examine not only the magnitude of APAs, but also the timing of APAs. If the MIP and MAP are separated for different magnitudes, then these timing differences should also be investigated. However, to date, there has been no study which examined the effect of perturbation magnitudes on timing of APAs. Therefore, the purpose of the current study was to investigate timing characteristics of the APAs in response to large- and small-magnitude perturbations. Specifically, we examined the effect of perturbation magnitude on the duration of the APAs in both the initial burst and sustained phases, as well as the onset and offset of these phases. Methods {#s2} ======= Ethics statement ---------------- The study was approved by the local Human Research Ethics Board. All of the participants gave written, informed consent to participate. Participants ------------ Nine people with chronic stroke were recruited for the study. The inclusion criteria included: (1) age 18--65 years; (2) hemiparesis (dominant or non-dominant upper limb, and lower limb) for at least 1 year after stroke onset (hemiplegia \> 1 year; see Figure [1](#F1){ref-type="fig"} for a summary of the recruitment process and the participation criteria); (3) ability to walk independently for 10 m; (4) ability to understand verbal instructions and follow the instructions during the experiment. The exclusion criteria included: (1) uncontrolled diabetes, cardiac, or lung diseases; (2) uncontrolled high blood pressure; (3) inability to participate in the required experiments due to other health conditions. Nine healthy participants were recruited from the local community for comparison purposes. The inclusion criteria for the healthy participants were: (1) age 18--65 years; (2) no past history of stroke or head injury; (3) ability to walk independently for 10 m; (4) ability to understand verbal instructions and follow the instructions during the experiment. The exclusion criteria for the healthy participants were: (1) uncontrolled diabetes, cardiac, or lung diseases; (2) uncontrolled high blood pressure; (3) unable to complete the required tasks. {#F1} Experimental protocol --------------------- The participants sat on a chair with an adjustable height