The idea of a neuromuscular synergy involving independent amplitude commands to the agonist and antagonist musculature is quite an appealing part of Plamondon's theory. One question that might be raised relates to the relative independence of the two “D” commands. Evidence is presented that suggests that the two commands might be related in sequential or simultaneous rapid aiming movements.

We question whether Plamondon & Alimi's model is useful in accounting for the nonsymmetrical and multiple-peaked velocity profiles observed in young and elderly subjects for ballistic aiming tasks. For these subjects, both data and observation suggest that a central representation initiates the movement in an appropriate direction but that multiple adjustments are made, both early and late, to achieve spatial accuracy.

Plamondon & Alimi propose a universal account of trajectory formation and speed/accuracy trade-off in rapid movements but fail, because: (1) the kinematic model ignores the more fundamental dynamics of movement generation, and (2) it does not capture the essential space-time constraints of movement accuracy. Hence, the modeling lacks a biologically and behaviorally principled foundation and is driven by pragmatic function fitting.

Even if we recognize that the delta-lognormal model provides an excellent fit to a large variety of data, the question remains as to what we actually learn from such a model, which could be seen as merely another multiparameter account? Do we welcome such an encompassing account, or do we expect to learn more from the limitations that become apparent when applying dedicated models addressing specific classes of movements?

Information about positions, from which differences in position are computed (as proposed in the vector-integration- to-endpoint model), provides a more plausible perceptual basis for the control of goal-directed arm movements than information about distance (as proposed in the kinematic model).

Although it provides a useful description of elementary movement trajectories, we argue that the delta-lognormal model is deficient as an account of speed/accuracy tradeoffs in aimed movements. It fails in this regard because (1) it is deterministic, (2) its formulation ignores critical task elements, and (3) it fails to account for the corrective role of submovements.

Several issues are raised concerning the notion that a single strategy explains Fitts' law and the linear speed/accuracy trade-off. Two additional concerns are discussed: (1) distance is programmed, (2) the fact that movements produced without the aid of vision obey Fitts' law does not mean that sighted movements must be explained without regard to vision.

To cover as much as possible the various questions raised by the commentators, I have divided my Response into three major sections. In section R1, I reply to the major comments and remarks dealing with the basic hypothesis upon which the kinematic theory is built (Plamondon 1993b; 1993c; 1995a; 1995b). I focus on linearity, determinism, kinematics, and the biological significance of the model parameters. I conclude this section by showing how, from a practical point of view, the delta-lognormal law can be used to group similar data prior to proceeding with further statistical analysis. In section R2, I address the main comments and remarks dealing with the speed/accuracy tradeoffs. First, I focus on the origin of speed/accuracy tradeoffs and the effect of visual feedback. Then I clarify some terminology problems and mathematical misinterpretations prior to providing new support for the theory using the experiments and data referred to by some commentators. I devote section R3 to a generalization of the kinematic theory and to some potential applications. I first provide new explanations of some classical experiments; then I show how more complex movements can be analyzed using the delta-lognormal law. I focus on perturbed and oscillatory movements as well as on cursive script and signature analysis. I conclude this section and my Response by showing some possible applications to the study of movement learning in children as well as to the study of aging phenomena in movement control. I complete my Response by recalling some technical problems that still need further analysis. Throughout the text, I try to convince the reader that the kinematic theory provides a new window on old and new data in the field.

I question (1) whether Karmiloff-Smith's (1994a,r) criticisms of modularity hit the target and (2) how much better the representational redescription model is. In both cases, “the mystery of the cognitive clock” is problematic for her account.

Representational redescription (Karmiloff-Smith 1994a; 1994) translates implicit, procedural knowledge into explicit, declarative knowledge. Explicit knowledge is an enabling condition of cognitive flexibility. The articulation and inferential integration of knowledge are important in explaining flexibility. There is an interesting connection to the availability of knowledge for verbal report, but no clear explanatory work is done by the idea of knowledge that is available to consciousness.

There are a number of elements in the representational redescription (RR) theory which elude definition, including behavioural success, implicit information, endogenous metaprocesses, and the detail of the representational levels. This commentary proposes an information processing approach to the development of cognitive flexibility – the Recursive Re-Representation (3Rs) model (Spensley 1995) – which redefines the developmental process and thereby eliminates these problematic concepts.

I respond to three continuing commentaries on Beyond modularity, two concerning the representational redescription (RR) framework and its attempts to account for the growing flexibility of human intelligence, and one relating to the putative mysteries of developmental timing. I discuss misunderstandings about the RR framework as well as some of its shortcomings. I strongly reject the notion of a genetic clock and go on to argue for epigenetic outcomes in which genes and environment interact during the protracted period of postnatal brain development.

Davey presents expectancy bias as an alternative explanation for preparedness effects in conditioning; this commentary mentions some of the problems in interpreting UCS expectancy designs vis-à-vis conditioning. A further complication is that other CS characteristics also influence conditioning: A recent experiment provided support for Rescorla and Wagner's (1972) hypothesis that the amount of conditioning is smaller when the CS and UCS elicit similarly valenced initial affective responses (as with a fear-relevant CS and shock UCS). Finally, different physiological systems measure different aspects of the conditioning process, a further consideration in interpreting the results of selective association studies.

The nonrandom distribution of fears is not as clearly related to phylogenetically survival relevance as preparedness theory seems to imply. Although delayed extinction reflects some of the best human evidence for preparedness, even this phenomenon is not as robust as it once seemed to be. Apart from the evidence reviewed by Davey, recent studies from our laboratory provide further evidence for an expectancy bias model of selective associations.

A series of arguments are presented by De Jong & Merckelbach which suggest that biological preparedness has been received significantly less critically than it should have been. I agree fully with their assessment. Cuthbert raises four questions about the applicability of the expectancy bias hypothesis to selective associations in human conditioning. This response argues that none of these four examples is necessarily problematic for the hypothesis.

An intertemporal bargaining model of commitment does not entail the interaction of parts within the person as Rachlin claims, and is needed to explain properties of “ordinary” self-control that his molar generalization model does not predict.

Rachlin's “teleological behaviorism” is a dubious melange. Of Aristotle's four basic “causes” – formal, efficient, material, and final – the scientists and philosophers of the modern era expelled the last, or teleology, from science. Adaptionist evolutionary biologists now sometimes sanction talk of the function or purpose of organisms' structures and behavioral repertoires as a first step because they believe evolution through natural selection makes natural organisms look as if they are purposively designed. But, as Aristotle himself insisted, humans are as much artificial as natural and so teleology is much less appropriate. To the degree that Rachlin's view makes sense it seems to amount to Daniel Dennett's intentional stance or the folk psychology talk of our everyday narrations of ourselves and others.

In response to Ainslie & Gault: The value of a temporally extended behavioral pattern depends on relationships inherent in the pattern itself. It is not possible to express that value as the simple sum of the discounted present values of the pattern's component acts.

In response to Leiber: Teleological behaviorism may be deemed unscientific because it has not yet succeeded to the required degree in predicting and controlling the highly complex patterns of human behavior that comprise our mental lives. However teleological behaviorism is not unscientific because it is teleological or “noncausal;” nor is teleological behaviorism unscientific because it is not reducible to neurophysiology. Nothing in principle bars the development of a teleological science of the mind.