Assessing the representation

Historically, motor performance has been described at the level of motor execution to examine accuracy (e.g., absolute error, root-mean-square error), kinematic (e.g., velocity profiles), and physiologic (e.g., electromyography) performance (Schmidt & Lee, 2005). These variables are useful in providing insight into general performance curves (e.g., A. Newell & Rosenbloom, 1981), as well understanding environmental and training variables affecting learning (Schmidt & Lee, 2005). Yet, performance data derived from the execution level of the motor hierarchy cannot provide direct information about the memory representation directing the motor response. Based on the general model of motor performance presented in Section 4.2.1, this memory representation is stored and selected farther up the hierarchy. The representation is influenced by a variety of factors operating at each level of the hierarchy (D. E. Meyer & Gordon, 1985; Rogers & Storkel, 1998; Rosenbaum, 1990; Schmidt & Wrisberg, 2004; Schmidt, 1988), such as variable mapping conditions between the response and stimuli (affecting response selection), accuracy demands (affecting response programming), and biomechanical constraints (affecting execution). Historically, these factors have been experimentally manipulated and the effects examined using reaction time (Klapp, 1996; Magill, 2001).

In the motor literature, reaction time has been described as the interval of time between the offset of stimulus presentation and the beginning of the motor response (Klapp, 1996; Magill,

movement or execution of the response, which is a separate measurment. The time from initiation of the movement until its completion is the movement time of the response (Magill, 2001; Rosenbaum, 1980). Thus, the overall response time is the reaction time plus the movement time (Magill, 2001), see Figure 8.

Reaction time and movement time are independent measures (Henry, 1961; Magill, 2001), and each can be mapped onto hierarchical information-processing models of motor control. Response selection and programming can be probed by reaction time, while execution of the movement can be assessed by movement time (Klapp, 1995, 1996; Magill, 2001). Movement time can be analyzed further using a fractionated reaction time. Premotor reaction time is defined from the initiation of electromyographic (EMG) activity to movement onset, and motor reaction time extends from the movement onset to its completion (Klapp, 1996; Maas &

Warning Signal “Go” Signal Initiation of the Response Foreperiod Reaction Time (RT) Response Time Termination of the Response Time  Movement Time (MT)

Figure 8: Overview of response time

Note: Adapted from Motor Learning: Concepts and Applications (6th ed.), Magill, pg. 25, 2001, with permission from McGraw-Hill Companies, Inc.

Mailend, 2012; Ono, 1990). Assessments of movement time have been employed to study the effects of response dynamics on goal-directed movement (Carlton, Carlton, & Newell, 1987; Klapp, 1996), as well as attention strategies on movement efficiency (Ono, 1990).

Although assessment of reaction time may describe both response selection and response programming, different variables interact with each stage (Klapp, 1995, 2003). Increasing response uncertainty, i.e., the difficulty in selecting the appropriate response, will increase reaction time during the selection stage of motor performance (Klapp, 1995, 1996, 2003). Response uncertainty can be manipulated by decreasing the compatibility, or learned association, between the stimulus and a given response (e.g., Fitts & Seeger, 1953). Other variables increasing response uncertainty during selection include: increasing the number of possible alternative responses (e.g., Brainard, Irby, Fitts, & Alluisi, 1962; Hick, 1952; Miller & Ulrich, 1998), varying the probability of competing responses (e.g., Hyman, 1953; Kornblum, 1969, 1973), and varying the number of parameters or distinctive markers that would distinguish between two alternative responses (e.g., Heuer, 1982; Rosenbaum, 1980, 1990).

Historically, motor theorists have only examined reaction time at the level of response programming (e.g., Henry & Rogers, 1960; Klapp, 1995; Osman, Kornblum, & Meyer, 1990; Sternberg et al., 1978), which assumes selection of the memory representation has already occurred. Slowed reaction times during programming are associated with increased complexity of the motor response. Complexity may be manipulated in a variety of ways, including the number of elements within a sequence (e.g., Anson, 1982; Canic & Franks, 1989; Christina & Rose, 1985; Deger & Ziegler, 2002; Henry & Rogers, 1960; Sternberg et al., 1978), the number of effectors (e.g., Schmidt & Wrisberg, 2004), accuracy demands (e.g., Sidaway, Sekiya, & Fairweather, 1995), and movement durations (e.g., Schmidt & Wrisberg, 2004). Each of these

factors requires an increase in organization and processing to program the sequence of movements, which results in longer reaction times (Klapp, 1996; Schmidt & Wrisberg, 2004). Complex motor behaviors may involve multiple motor representations, which are sequenced in a “buffer,” or cognitive space (Klapp, 1996; Osman et al., 1990; Rogers & Storkel, 1999; Verwey, 1995). The nature of the buffers varies based on the motor performance model, with some models preferring capacity limitations constraining their size (e.g., Rogers & Storkel, 1999) and other placing no restrictions on the capacity of the buffer (e.g., Deger & Ziegler, 2002). Thus, reaction times assessed at the programming level are influenced by the memory representation, interactions with complexity variables, and potential buffer capacity limitations.

As noted earlier, the information-processing models emphasized here are hierarchical, i.e., processing at one level will have an impact on subsequent stages (D. E. Meyer & Gordon, 1985; Rogers & Storkel, 1998; Rosenbaum, 1990; Schmidt & Wrisberg, 2004; Schmidt, 1988). Therefore, reaction times assessed at the level of programming may have been influenced by higher-level processing, e.g., at the levels of selection and/or cognitive input stages. Many motor theorists have attempted to assess memory representations by manipulating programming variables and recording reaction times (e.g., Chamberlin & Magill, 1992b; Crump & Logan, 2010; Gordon & Meyer, 1984; Ludlow, Connor, & Bassich, 1987); yet, a true assessment of memory representation cannot be evaluated at the level of programming. The memory representation for the motor behavior would have already been selected (i.e., processed at the response selection stage) and manipulated during the programming stage. Investigators attempting to study the memory representation of motor control via reaction time, therefore, need to access the response selection stage without the cumulative confounds of the programming or execution stages influencing the reaction time.