Dear Andrew,
I'll respond first to the "specificity of practice" hypothesis and then to ProAim issues.
In general, my experience has supported the basic notion that the modality relied upon in practice for so-called "afferent cues" tends to dominate performance. In plain English, if you learn by the eyes, you die by the eyes; and if you learn by kinesthetic proprioception, you die by the same. But I only agree with this in general, and that is because sports science has a rather major problem at its core: task definition.
The thrust of all standard academic sports science studies is to test an hypothesis against empirical investigation of a specific task, and then make generalizations about human performance based upon the data and whether it appears to support or contradict the hypothesis under examination. The devil in the details here is in defining and understanding the task. Without clear definitions of the task and what sorts of behaviors it calls upon for its performance or the variables at work in its execution, it is problematic in the extreme to interpret the data and make sound, useful conclusions about human performance regardless of the rigor of the methodology. Sports science has always suffered from something of a willful blindness to this central difficulty, mostly I suppose because it is a damned inconvenient nuisance.
Let's take the work of Luc Proteau at the University of Montreal's Department of Kinesiology as an example, because his work on this specificity of practice hypothesis pretty much leads the field. In his more recent work on this topic,
L. Proteau and H. Carnahan, What Causes Specificity of Practice in a Manual Aiming Movement: Vision Dominance or Transformation Errors? Journal of Motor Behavior 33(3) (Sep 2001): 226-234, the investigators use a bimanual arm pointing task to attempt to find out whether withdrawal of vision of the arm during pointing at a target causes large errors in accuracy because the pointing is visually learned, or the withdrawal of vision requires a translation of targeting into body-centered rather than visual coding and this process is faulty, or both. The subjects learn the arm pointing task with one hand and then the other using vision of the target and hand or using vision of the target alone, and then are tested in performance with vision of the hand removed. The authors conclude that the large errors in performance are due to both reasons.
In the course of reviewing the state of the scientific literature up to date, the authors recount the findings in the 1998 study you cite: Tremblay, L., & Proteau, L. (1998). Specificity of practice: The case of powerlifting. Research Quarterly for Exercise and Sport, 69, 284-289. Because the notion of "specificity of practice" is probably not well understood by visitors to this site, allow me to quote at length from the 2001 article, as these passages crystalize the issues fairly well. Let me say up front that "afferent information" is jargon for "sensations," either visual or physical. Key passages are made bold face:
"The role played by afferent information, and especially by visual information, in the control of movement has been the subject of debate over the years. At one extreme of the continuum, some investigators have proposed that the role of afferent information for movement control is limited to the early stages of learning and that the processing of afferent information plays a very minor role in experts' performance (Keele, 1968; Schmidt, 1975). At the other end of the spectrum, others have proposed that afferent information might become more important for motor control as practice increases (see Proteau, 1992, for a review). In addition, the advocates of the view that afferent information plays an important role in movement control do not agree on how that information is used as a function of practice. On the one hand, some authors have proposed that the source of afferent information used to guide one's action differs early and late in practice. For instance, Fleishman and Rich (1963; see also Bennett & Davids, 1995) suggested that visual afferent information is the main input for movement control early in practice but that it is gradually replaced by the use of proprioceptive information. On the other hand, Tremblay and Proteau (1998) recently proposed that, regardless of the task one practices, there is a single source of afferent information that is quickly determined as being best suited to ensure optimal accuracy, and that source progressively dominates other sources of afferent information for motor control. It is possible that because one becomes more proficient at processing the primary source of feedback as a result of practice, progressively optimal performance is ensured (see Abrams & Pratt, 1993, and Khan, Franks, & Goodman, 1998, for confirming evidence).
A corollary of the point of view developed by Tremblay and Proteau (1998) is that learning must be specific to the sources of afferent feedback used to guide one's movement during practice. Support for this specificity of practice hypothesis has been obtained in numerous studies of aiming movements (see Proteau, 1992, for a review of early work). For example, Proteau, Marteniuk, Girouard, and Dugas (1987) had participants practice an aiming movement for 200 or 2,000 trials under either a full-vision condition or a condition in which they could see only the target to be reached (target-only condition). Not surprisingly, by the end of practice, participants who had practiced under the full-vision condition were more accurate than participants who had trained under the target-only condition. More interesting results were obtained in a transfer test performed immediately after the last practice trial and performed by all participants in the target-only condition with no feedback about the end result of their movement. The results of the transfer test revealed a huge increase in error for participants who had trained in the full-vision condition, who, regardless of the amount of practice, became significantly less accurate than participants who practiced in the target-only condition. Moreover, the increase in error was even larger for participants who had trained for 2,000 trials than for those who had trained for 200 trials.( n1) Thus, learning appears to be specific to the sources of afferent information used to ensure optimal accuracy during practice; and when that information is withdrawn in a transfer test, performance suffers because the individual has no reliable source of reference with which to evaluate his or her movement.
There is, however, an alternative--although not mutually exclusive--explanation of the findings just described. It has been proposed that when practice occurs in a normal visual context, both the target and the effector are coded within an allocentric (visual) frame of reference. Thus, errors in the ongoing movement can easily be detected and corrected. When vision of the effector is not permitted, however, the location of the target is still coded within an allocentric frame of reference, whereas the location of the effector is coded within an egocentric (proprioceptive) frame of reference (Jeannerod, 1988). The egocentric frame of reference is thought to be anchored on the sagittal midline of an individual's body (Bartolemeo & Chokron, 1999). It is thought that, for movement to occur, participants must transform the location of the target from its allocentric frame of reference to the egocentric frame of reference, in which the effector is coded. The larger pointing errors found in a context in which only the target is visible have been attributed to transformation errors that occur when the representation of the target is transformed from its allocentric frame of reference to the egocentric frame of reference (Soechting & Flanders, 1989a, 1989b; Vindras & Viviani, 1998). Thus, it is possible that the large increase in errors noted in the transfer tests of the studies just reviewed simply reflects errors resulting from transformation from one frame of reference to another, rather than indicating that learning is specific to the sources of afferent information used to ensure optimal accuracy during practice. Our goal in the present study was to dissociate those two possible explanations.
In the present study, we used a bilateral transfer paradigm. Specifically, we had participants practice a manual aiming movement with either their left or their right arm under a normal vision condition or under a target-only condition. During the training phase, it is hypothesized, participants in the target-only group will learn to reduce transformation errors occurring because the target and their arm are not coded in the same frame of reference, which will result in a decrease in spatial aiming error. During that phase, participants in the normal vision group should be more accurate than those in the target-only group because normal vision allows them to code both the target and their arm in the same frame of reference. Following practice, all participants performed transfer trials in the target-only condition, using both their trained and their untrained arms. If the large number of errors found in the transfer phases of previous work (Proteau, 1992) reflects only transformation errors concerning the location of the target, withdrawing vision of the effector in transfer should have similar effects for both the trained and the untrained arms (large increase in spatial error). That is so because, regardless of the arm one uses, the location of the target has to be recoded in the egocentric frame of reference, which will result in transformation errors. In addition, the specificity of practice hypothesis should be rejected if smaller aiming errors are found for the trained arm than for the untrained arm. That conclusion can be drawn because in the specificity of practice hypothesis it is posited that the availability of on-line visual afferent information masks the processing of other sources of afferent information (Tremblay & Proteau, 1998). If it does so, then withdrawing vision in transfer should leave the participant without optimally developed proprioceptive reference mechanisms for the trained arm. At best, it should result in aiming errors for the trained arm as large as those found for the untrained arm, for which such a proprioceptive reference has not been developed because that arm has not been used in practice."
So what are they saying? As I undertsand them, the "specificity of training" hypothesis means that one modality (e.g., vision) asserts itself in practice as the dominant modality, so that visual cues dominate the learning process; that this dominance had a suppressive effect on learning in the other modality (e.g., proprioceptive or physical-motion learning); that subsequent withdrawal of vision reveals this lack of trained competence in the non-dominant modality; and that optimizing performance requires training with the dominant modality and then performing with the dominant modality.
I don't fully agree with this in the context of putting, because of the nature of the task. The only MOVEMENT in putting is always the same, except for amplitude. There is not really a "pointing" or "aiming" MOVEMENT in the task of putting as it is normally understood. Although vision clearly dominates early learning, I don't agree that this is the best modality for performance execution, although I agree it suppresses early learning and indeed longterm focus on proprioceptive and kinetic learning, which is more vital.
The object of putting is to aim the putter face at a target, setup to the putter face, and make a straight ("same every time" stroke) with good touch (amplitude). Once the putter face is aimed, it is time to perform the stroke. The stroke MOVEMENT itself is always the same, so there is not really any pointing or aiming in the movement itself. My experience has been that the natural dominance of vision not only masks body-motion learning, but that it conceals the basic truth that vision is almost utterly irrelevant to the motion of the stroke per se. It is only once that the limited role of vision in the straight same-every-time stroke is realized that the required body-motion skills come to the fore and receive the appropriate level of attention and training. I suppose vision retains a rudimentary value in assisting the same-every-time stroke motion, but it is not really all that helpful. At the end of the day, the body's motion is more important to accurate execution, although vision helps a little so long as it stays in a supporting role.
The problem of sports science in general is that the experiments and data gathering always depend upon very specific tasks, but these tasks may or may not be well understood or even appropriate for application in generalized ways. In neuroscience, it has long been recognized that the functional nuances of the task are in fact the real quarry to understand, such as the difference between hearing noises, recognizing distinct cultural phonemes, comprehending the sounds of individual words, resolving the ambiguity of meaning of the word as used contextually and with intonations, and understanding the syntax and semantics of a sentence. Each of these is a separate function in the brain, involving separate neural circuitry and learned association patterns. Sports scientists, in contrast, are usually too ready to interpret experiments based upon a conveniently-devised task and then to generalize from that task-specific data to widely-applicable statements about the best way to learn and perform "skills". But for a performer to USE this vast body of sports science in application to a specific task like putting requires another layer of analysis and "fitting" of the experimental data to the putting task, separate and apart from anything the sports scientist may have concluded.
This difficulty is often highlighted in so-called "metastudies" of sports science literature, where authors of an overview of a body of related scientific studies scrutinize the tasks and the methodology in an attempt to reconcile and integrate diverse studies and differing data sets. Generally, these studies get graded by how well they define the tasks and the environmental influences subject to experimental control and isolation, and also by the appropriateness of the tasks as a foundation for generalization to skills. The skills and their component tasks central to these studies are generally defined based upon how the experimenter understands so-called "elite performance" normally represented by the "expert model." The pretense, sometimes amusing, is that academic sports scientists actually have a deep appreciation for the skills they design studies to examine. Sometimes they do, but by far the more common experience is for the academic to start out with some half-cooked generalizations about skilled performance based on having perceived some sort of consensus view in the literature of the sport based on a cursory and superficial examination of sources.
Basically, I haven't seen much evidence that sports scientists have a good, detailed grasp of the functions at work in putting skills, in any of the sensory or motor modalities. I've attempted on a number of occasions to communicate with folks like Debbie Crews (sports psychology) at Arizona State, or Bob Christina (motor learning) at UNCG, or Joan Vickers (sports vision) at the University of Calgary, or Aynsly Smith (sports psychology) at the Mayo Clinic, or Betsy Clark (spots education) at the LPGA, or any number of other academics, but they just don't respond. The same is true with the work of Luc Proteau (sports kinesiology). The basic tasks he and others study are not really applicable to the task of putting, at least in the way I undertsand it. Putting is not, in an optimal way, a matter of "pointing" or "reaching" to a target (
pace, Eben Dennis), and that sort of motor control recruits a whole set of neural processes not applicable to making a stroke. The stroke I teach, which is a straight stroke, really has "dead hands" and no arm action, and consists in nothing more than moving the shoulders with the gut muscles. There IS no transformation of the effector (shoulder motor control) from allocentric to egocentric coding in the putting stroke: once the putter face is aimed, the body sets up square to the putter face; once that is done, the golfer just putts the same as always reagrdless of target location in any coordinate frame.
So, in summary, while I don't agree that whatever emerges early on as the dominant modality is also the optimal modality, I find the specificity of practice hypothesis generally valid in most respects, but the problem is in the task being practiced. Unless this is defined better, the application of the theory doesn't make much sense. And in any event, I believe that each modality has to be trained, and that overtraining the lesser modality at the beginning does not prevent fully training the other modality, even if it delays it. To the extent the "specificity of practice" hypothesis creates a performance problem by dint of inattention to optimal training modality, the problem is scarcely intractable.
With regard to the ProAim goggles, I believe these do exacerbate the basic problem of training a same-every-time stroke with body-motion as if it were a problem of matching hand coordination to a visual channel. The action of the shoulders, with dead hands and a set triangle shape, is sufficient in itself to deliver the putter head on line back and thru. There is no occasion to recruit the arms and hands in guiding the putter head thru or along a visual channel. So, yes, the ProAim goggles lead golfers down the wrong path and in fact make it harder to train the body action. The stroke motion does not need vision except in a limited supplementary way, and the ProAim glasses encourage golfers to think otherwise, train as if the "visual channel' settles all issues, and ignore the more vital proprioceptive and kinetic training (the "feel" of the stroke).
On the other hand, the ProAim goggles serve a useful function for the "task" of setting the body square to the putter face, as well as the "task" of checking the putter face's aim as set based upon the "task" of targeting from behind the ball and walking into the putt to place the putter face. That's because the skull line needs to be square to the putter face and the eyeballs gazing straight out for these two post-face-aiming "tasks" to be performed well. But after that, proprioceptive kinetics takes the dominant role in the stroke, with vision only stabilizing and supporting the movement control.
Cheers!
Geoff Mangum
Putting Theorist and Instructor
Geoff Mangum's PuttingZone
Golf's most advanced and comprehensive putting instruction.
Over 645,000 visits and growing strong ...
518 Woodlawn Ave
Greensboro NC 27401
336.230.0612 home
336.402.1602 cell
