Table 16.1. Summary of principles of sensory training derived from theories of perceptual learning, neural plasticity and physiology of the somatosensory system.
Derived from and/or consistent with theories of
Principle of training
Repeated stimulation of specific stimuli Attentive exploration
meaningful task Use of anticipation trials (and imagery)
- method of exploration and summary feedback
Calibration of sensation:
within and across modalities Progression from easy to more difficult discriminations
Variation in stimuli, intermittent feedback and tuition of training principles
7 improvement maximal for specific stimulus 7 attention important in learning
7 link with attention, learning and success
7 extrinsic and summary feedback may enhance learning
7 improved perception
7 required for learning within and across sensory dimensions 7 improved performance and retention 7 facilitates transfer of training effects
7 modification of sensory map is task-dependent 7 modulation of neural plasticity 7 force use of somatosensory system 7 brain responds to meaningful goals 7 activation of similar sites as for direct stimulation. 7 enhance new connections
7 cross-modal plasticity
7 progressively challenge system
7 forced use, competitive use
7 specificity of processing within the system 7 component of sensory processing 7 vision may dominate
7 consistent with normal processing
Most of the principles identified from these fields are complementary, consistent with the knowledge that neural plasticity underlies learning and recovery following injury. In addition, the physiology of processing somatosensory information must be considered in application to the sensory system. More recent sensory retraining approaches have begun to base their training on these principles (see Section Review of documented programs in relation to basic science and empirical foundations of this chapter for review). Key principles that have been successfully applied, or have potential for application, are discussed below. Their application is summarized in Table 16.1.
Neural plastic changes are experience-dependent (see Section A on Neural plasticity in Volume I; Nudo et al., 1996a) and the system is competitive (Merzenich and Jenkins, 1993; Weinreich and Armentrout, 1995). Functional reorganization has been demonstrated after behaviorally controlled tactile stimulation in intact animals (Recanzone et al., 1992). Forced use of the system, for example using constraint induced movement therapy, has also been associated with neural plastic changes in post-stroke motor recovery (Liepert et al., 2000). Even less intensive programs based on principles of motor learning found training-induced changes (Carey J.R. et al., 2002). Importantly, repetitive use alone may not be sufficient to effect changes in cortical representation. Rather changes are associated with specific skill learning, consistent with a
"learning-dependent" hypothesis of neural plasticity (Karni et al., 1995; Plautz et al., 2000). Sensory neurorehabilitation should therefore challenge the sensory system using repeated stimulation of targeted sensory tasks coupled with an intensive perceptual-learning based training program.
Learning is reported to be maximal for the specific task trained (Gibson, 1969; Goldstone, 1998; Sathian and Zangaladze, 1997), consistent with highly specific organization of the system (see Section Definition and processing within the somatosen-sory system of this chapter for review) and evidence that functional reorganization is directly linked with changes in cortical regions engaged by these inputs (Recanzone et al., 1992). Repetition over time with an increasing number of coincident events serves to strengthen synaptic connections (Byl and Merzenich, 2000). However, competition between afferent inputs for connections in the sensory cortex (Merzenich and Jenkins, 1993) suggests the need for caution with overstimulation of a specific site at the expense of related areas. Consequently to maximize task-specific learning, training stimuli and the method of processing the information should match targeted discrimination tasks (Carey et al., 1993). Training of important sites normally responsible for the sensation, for example the hand (Dannenbaum and Dykes, 1988; Yekutiel and Guttman, 1993), and discriminations characteristically impaired and important for daily function (Carey et al., 1993) have been recommended.
Motivation is important in learning (Goldstone, 1998) and recovery after brain injury (Bach-y-Rita, 1980), and the brain responds to meaningful goals (Nudo et al., 1996b). Training should therefore be goal directed, interesting and demanding if it is to tap into the brain's potential for functional reorganization (Byl and Merzenich, 2000; Yekutiel, 2000). It should provide regular opportunities for success, with reinforcement to encourage motivation and participation (Carey, 1993; Yekutiel, 2000).
Attention is crucial in perceptual learning. Attentive exploration of stimuli allows purposeful feedback within the sensory-perceptual system (Epstein et al., 1989) and perception becomes adapted to tasks by increasing the attention paid to important features (with less noticing of irrelevancies) (Gibson, 1969; Goldstone, 1998). Perceptual learning (Goldstone, 1998) and neurophysiological (Johnson and Hsiao, 1992) evidence propose that "distinctive features of difference" are learned and form the basis of transfer of training. Attention is also important in the modulation of cortical plasticity (see Chapter 12 of Volume I), particularly in early stages of plastic changes and learning (Karni et al., 1995). Thus patients should actively (where possible) and purposively explore sensory stimuli with attention directed to distinctive features of difference. This may be facilitated through requiring a response and guiding the patient to search for distinctive features (Carey et al., 1993). Further, as vision may dominate tactile and proprioceptive senses in some instances (Clark and Horch, 1986; Lederman et al., 1986), exploration of stimuli with vision occluded should be included to allow subjects to focus specifically on the somatic sensations (Carey et al., 1993).
Anticipation may facilitate recruitment of existing or new sensory sites in the brain. Similar brain sites are active under direct stimulation and anticipated stimulation conditions (Roland, 1981). Further, prior and subjective experience can influence early stages of information processing and facilitate stimulus differentiation (Goldstone, 1998). Anticipation trials, in which the patient is informed that a limited set of previously experienced stimuli will be used (Carey et al., 1993), may tap into this capacity and encourage new neural connections. A patient may also be encouraged to imagine what a stimulus should feel like, based on evidence that haptic (tactile and proprioceptive) information is represented through imagery (Klatzky et al., 1991).
Augmented feedback on accuracy of response outcome and on performance is important in skill acquisition (Schmidt and Lee, 1999) and may enhance perceptual learning (Gibson, 1969). Although improvement in perceptual discriminations may be experienced in unimpaired subjects without extrinsic feedback in some cases (Epstein et al., 1989), practice with correction can enhance learning
(Gibson, 1969). We found that repeated exposure alone was insufficient to effect a positive training outcome in the majority of cases post-stroke (Carey et al., 1993; Carey and Matyas, 2005). Thus feedback on accuracy of response and performance, for example method of exploration (Lederman and Klatzky, 1993), should be provided. Feedback should be immediate, precise and quantitative to maximize acquisition (Salmoni et al., 1984). Summary feedback also enhances learning (Salmoni et al., 1984) and should be provided at the end of each training session.
Calibration of perceptions would also appear to be important. This may involve comparison of the sensation with the other hand (Gibson, 1969) and use of vision to facilitate cross-modal calibration (Lederman et al., 1986), consistent with activity in visual cortical regions during tactile perception (Zangaladze et al., 1999). Computational neural models indicate that when two modalities are trained at the same time and provide feedback for each other, a higher level of performance is possible than if they remained independent (Becker, 1996). Moreover, cross-modal plasticity in sensory systems (see Chapter 11 of Volume I) may facilitate alternate and new neural connections.
Finer perceptual differences are able to be distinguished through exposure to a series of graded stimuli (Ahissar and Hochstein, 1997; Goldstone, 1998). Graded progression facilitates perceptual differentiation, especially of complex stimuli, as presentation of an easy discrimination first allows the subject to allocate attention to the relevant dimension (Goldstone, 1998). Further, transfer from one stimulus to another within a unidimensional sensory quality requires presentation in a graded manner (Gibson, 1969). Thus training should progress from easy to more difficult discriminations across stimuli and within a unidimensional sensory quality.
Performance is better on frequently presented items than rare items (Allen and Brooks, 1991). In addition, best available evidence suggests that training should continue for some time after "mastery" to increase retention (Lane, 1987). This further suggests the need for repetition and intensive training.
Specificity of learning and principles to facilitate learning transfer
Whilst the above principles have been associated with positive perceptual learning, there are limits on the generality of perceptual learning. Perceptual learning is usually highly specific to the task, receptor location and method of processing (Sathian and Zangaladze, 1997; Goldstone, 1998). Similarly, we found highly specific training effects in tasks employing the same sensory dimension (tactile or proprioceptive), sub-modality (grid or fabric textures) and body location (fingertip or wrist) with SST post-stroke (Carey et al., 1993; Carey and Matyas, 2005). However, perceptual transfer is possible in some instances with unimpaired subjects (Epstein et al., 1989; Ettlinger and Wilson, 1990; Spengler et al., 1997). Similarity between original and transfer tasks is an important factor influencing the degree of transfer (Gibson, 1969). It has been suggested that transfer should be more prominent where the stimuli are more complex and potentially share a number of distinctive features (Gibson, 1991; Goldstone, 1998). Transfer across body sites has also been reported in unimpaired systems (Sathian and Zangaladze, 1997; Spengler et al., 1997), and may be influenced by the attention demands and complexity of the task (Ahissar and Hochstein, 1997).
Principles of learning that facilitate transfer of training effects in unimpaired subjects have potential application following injury. Transfer of training effects is more effective when variation in stimuli is employed (Gibson, 1991; Goldstone, 1998; Schmidt and Lee, 1999). Optimally this should include training across a variety of stimuli with a wide range of distinctive features, for example roughness characteristics, as well as variation in tasks and environments (Carey and Matyas, 2005). To achieve grading, progressive difficulty should be defined across stimuli sets as well as within sets (Carey and Matyas, 2005).
Intermittent feedback on accuracy of response (Winstein and Schmidt, 1990) and specific instruction on principles of training and how these apply across tasks (Cormier and Hagman, 1987) have also been associated with enhanced transfer and retention. Further, an important part of learning transfer tasks is acquiring the capacity to cope with novel situations (Schmidt and Lee, 1999). This suggests the need to provide exposure to novel stimuli with opportunity to get feedback on the act of generalization (Carey and Matyas, 2005).
In summary, positive findings from studies of sensory retraining (see section Review of documented programs in relation to basic science and empirical foundations of this chapter for review) support the application of principles of training derived from literature on perceptual learning and neural plasticity. Further, the nature of training, that is stimulus specific versus generalization optimized, appears crucial to outcome. Evidence of a learning phenomenon associated with sensory retraining post-stroke has been quantified in the intervention time-series data of our patients (Carey, 1993; see also Fig. 16.2). The improvement curve was consistent with that described in the learning literature (Lane, 1987; Epstein et al., 1989) and in studies of neural plasticity (Recanzone et al., 1992). However, in contrast to unimpaired subjects (Epstein et al., 1989), the characteristic learning curve was only achieved under supervised training conditions (Carey et al., 1993). The potential for generalization of training within a sensory dimension has also been demonstrated post-stroke, provided a program designed to enhance transfer is used (Carey and Matyas, 2005). Other programs that have employed training across a variety of tasks have also found generalized training effects (Byl et al., 2003; Smania et al., 2003; Yekutiel and Guttman, 1993).
16.6 Future directions for the integration of basic science in clinical practice, as applied to neurorehabilitation of somatic sensation
Brain networks may reorganize to optimize stroke recovery. However, despite evidence of behavioral improvement, it is not known to what extent training-induced recovery is associated with changes in the functional neuroanatomy of sensation in humans. In particular it is unknown whether sites different to those typically used are involved in the recovery process, possibly suggestive of different behavioral strategies. Clinically effective sensory training programs, derived from theories of perceptual learning and recovery following brain damage, may be used to test for outcomes related to brain adaptation. Confirmation of the prediction that neural plastic changes occur primarily within the pre-existing somatosensory system (Weinreich and Armentrout, 1995), will highlight the importance of sparing and dynamic adaptation within pre-existing sensory sites. Evidence of recruitment of different sites may identify involvement of other systems important in recovery, such as attention (see Chapter 12 of Volume I) and visual (see Chapters 9 and 11 of Volume I) systems. This will advance our understanding of whether training is operating at a level of restitution or substitution within the system and provide insight into behavioral strategies associated with training. Comparison with motor recovery findings will help determine if there is a common model of neural plasticity associated with motor and somatosensory recovery.
Systematic investigation of the conditions under which behavioral improvement and neural repair is most effectively achieved is required to provide ongoing direction for the development of science-based interventions. For example, the timing (post-injury) and intensity of training requires investigation, given evidence of possible maladaptive changes (Merzenich and Jenkins, 1993; Nudo et al., 1996b). Investigation of the need for specific training, compared to nonspecific exposure, and the neural outcomes of learning transfer (Spengler et al., 1997) will help elucidate the mechanisms and brain regions involved in different training methods. Similarly, different principles of training (e.g., cross-modality matching and feedback on accuracy) require systematic investigation of their contribution as they may involve different neural structures and mechanisms. Paradigms that permit investigation of component stages in neural processing need to be conducted and interpreted using models of analysis that focus on connectivity within the system.
Knowledge of the relationship between brain activation and recovery will have predictive significance in relation to identifying patients who are likely to show spontaneous recovery and/or who are able to benefit from training. Potential explanations for individual differences in the nature and extent of recovery may relate to lesion site (Zemke et al., 2003) and remote changes in structure, including diaschi-sis (Seitz et al., 1999). Investigation of the association between structural brain changes and the brain's capacity for reorganization is indicated.
Further development of current training programs is also indicated. Programs reviewed (see Section Review of documented programs in relation to basic science and empirical foundations of this chapter for review) have incorporated principles consistent with perceptual learning and neural plasticity. However, individual features of the training protocols have not been dissected. This is necessary to identify the critically important elements associated with successful learning and transfer. Previous studies have provided some insight into the boundaries of spontaneous and facilitated transfer with stroke patients (Carey et al., 1993; Carey and Matyas, 2005). Investigation of similarity of trained and transfer tasks, method of information processing, level of task difficulty and relative body location on learning transfer is needed to guide clinical training programs and clarify the nature of what is being learnt.
The most optimal combination of principles may also vary with individual patient characteristics including the nature of loss (e.g., detection versus discrimination versus multisensory integration), the phase of recovery (acute versus chronic) and the site of lesion (PNS versus CNS or specific location within these). Systematic investigation of these is indicated. The focus of training could also be expanded to include training in discrimination of size, shape and weight of objects, detection of slip when holding objects, regulation of pressure during grasp and spontaneous use of the limb. Finally, the ability to modify sensory abilities experimentally opens up an experimental paradigm for future investigations of the relationship between sensation and other abilities, such as pinch grip, and the effect of improving sensation on motor function and activities of daily living.
Was this article helpful?