Abstracts

The biomechanical analysis concerns the elementary functional movement (third level of action following J.F.Richard, 1990), i.e. the movement performed by the craftsman between two subsequent percussions. This movement is naturally considered as "elementary" from the point of view of general strategies of stone-knapping. However, elementary movement is in turn quite complex from biomechanical point of view because of many degrees of freedom (DoF) contributing in the movement. It is natural to assume that all mechanical DoF are highly co-ordinated in the adaptive and dextrous skill. The biomechanical analysis consists in the revealing of kinematic and dynamic synergies of the elementary movement, i.e. of the co-ordination between the joint angles and joint torques respectively.

The human arm is considered as the system of three rigid bodies (the hand, the forearm, and the arm) connected by the frictionless joints (the wrist, the elbow and the shoulder). This system has seven mechanical DoF: two DoF in the wrist joint, corresponding to the abduction-adduction and flexion-extension of the hand relative to the forearm; two DoF in the elbow joint, corresponding to the pronation-supination and the flexion-extension of the forearm relative to the arm, and, finally, three DoF in the shoulder joint, corresponding to the abduction-adduction, flexion-extension and rotation of the arm relative to the body. The data from Fastrack system used for the movement recordings allow for calculation of joint angles corresponding to the above mentioned DoF (Biryukova et al., 2000). The time courses of these angles during the elementary movement completely describe its kinematics.

The joint torques have been calculated using the formalism described by Korenev (1979) and Biryukova & Yurovskaya (1994). The numerical derivatives of the calculated joint angles were used as enough accurate assessments of angular velocities and angular accelerations (Prokopenko et al., 2001). Inertial characteristics of the craftsmen arms were calculated on the base of direct measurements of the weight, the height, the lengths of arm segments and the volumes of craftsmen arms.

The co-ordinations between the joint angles and between the joint torques are investigated by the method of principal components (PC). The angular variance accounted by the first PC is accepted as a degree of correlation between the joint angles and joint torques. The loading factors showing the contribution of each joint into the elementary movement are accepted as the description of its kinematic and dynamic synergies. The procedure of PC analysis was applied to elementary movements corresponding to different sub-goals of basic strategy: 1) calibration of the crest; 2) end preparation; 3) crest fluting; 4) axial removals (Roux et al., 1995). The PC analysis was also applied to the knapping on the anvil (a sharp-pointed iron bar). Despite of these movements are not functional they are often made by the craftsmen during the stone knapping. The cluster analysis was used for comparative analysis of the enlisted movements.

The preliminary results of a biomechanical analysis are the following:

1) There exists the stereotypic kinematic pattern of elementary movement, which is similar for different sub-goals of the basic strategy. The rotations in the arm joints are highly correlated. These facts give us an idea that the control of the elementary movement can be fairly simple, despite of large number of DoFs involved in the task. This control can consist in one central command common for all joints. The correlation of movement amplitude with the contributions of joint angles into the movement allows supposing the linear scaling of this command for the control of elementary movements with different amplitudes. There are many evidences to assume that this type of control consists in the preset of visco-elastic parameters of the arm joints and in the signal for movement starting. It is therefore feed-forward control, which does not intervene during the movement. In this sense, the movement is executed in a "passive" manner.

The objective point of percussion varies significantly during the stone knapping. The accuracy of percussion can be achieved in two ways: by the flexibility of stereotypic kinematic pattern, and (or), by the appropriate installation of the stone by the postural hand. In the last case, the postural arm plays the main role in the task, leaving to the "knapping" arm to perform the "passive" stereotypic movements.

2) The movements of anvil knapping are often appeared during the stone knapping. One may assume that this is for "keeping in mind" the end of the anvil as crucial point of the task. The high variability of anvil knapping gives an idea of less strong control, which can mean the relaxation during these movements. The third explanation is that the craftsmen "remember" and "keep active" the repertoire of different movements giving the good result. This repertoire makes the motor automatism (Bernstein, 1947; 1991). We can hypothesize that the experts have a larger repertoire of elementary movements than the craftsmen of lower level of expertise.

Faculty of Human Movement sciences, VU, Amsterdam

The present study addresses changes in action possibilities with a tool and how those changes develop. We emphasize the importance of aspects related to action, instead of focusing on how children create new building blocks in an action plan to incorporate the tool. We study how actions are adapted to control the interface between tool and goal-object. We focus on the strategies children use to discover how to constrain the degrees of freedom in the action system so that the new ‘end-effector’, which is displaced from the body to the tool, becomes effective.

With a rod pointing upward, young children (aged two to four years) approached an object. They stopped, lowered the rod and displaced the object with the rod's tip. To be successful, the child has to stop at a place that both accommodates rod length and leaves room for an appropriate posture to control the rod. Rod length and mass, and object size were varied and all three variables were expected to affect control of the rod. For instance, a smaller object requires more movement precision. We examined whether and how those variables affected children's behavior in different phases of the performance (e.g., approach, stopping place, and how the rod was held). The required control of the rod should affect actions early in a trial and such early adaptations reflect whether action possibilities with a tool are perceived. Younger children lowered longer rods early, during the approach. Also, when displacing a smaller object, two hands were often used early in a trial. Rod mass affected the fine-tuning of the distance just before or during the displacement. The latter two effects did not vary with age.

The alterations in actions were functional with regard to the experimental manipulation, and the rod-lowering created information about action possibilities with the tool. This indicated that children adapted their actions to create information about how the tool changed their action possibilities. In other words, actions were adapted to learn how to control the tool with respect to the goal of the task. Possible implications for studies on stone-knapping will be discussed.

INSERM 483, Paris

Cambay is one of the very rare place in the world where the knapping stone technique still responds to the principles of the conchoïdal fracture (cf. abstract J. Pelegrin). The technique practised is an indirect percussion by counter-blow and is used for making stone beads. This technique is a recent one in the history of techniques. However, it can provide an appropriate reference situation to study the skills involved in stone knapping. In effect, given that for all stone knapping techniques, the stable parameters of conchoïdal fracture must be controlled, we can assume then that similar skills are involved.

In order to characterise the skills involved in stone knapping in Cambay, a program of field experimentation was developed. At Cambay, the knappers are distinguished according to different levels of competence. In psychology, studies of apprenticeship are often based on transfer tasks. We tested the following hypothesis: the degree of success attained in modified situations reflects the capacities of adjustment, flexibility and planning of the artisan. A highly skilled artisan should be capable of transferring his planning and motor skills to new situations. In addition, we would expect that the differences that characterise distinct expertise levels to be amplified in the context of transfer tasks. Modifications were introduced into the knapping activity of the artisan, varying the nature of the raw material to be worked (glass versus stone) and the objective to be attained (bead dimensions and shapes). We worked with 12 artisans distributed equally among two levels of expertise: a high level, group 1, averaging 44 years of age; and a low level, group 2, with an average age of 28 years. Each artisan was asked to knap a total of 80 beads. They were instructed to produce superior quality beads. The ensemble of fabrication processes of each bead was recorded with a video camera and an accelerometer attached to the head of the knapper’s hammer in order to analyse the succession of actions realised, as well as the characteristics of the elementary movements and their sequencing.

The four categories of data have been analysed: the finished products, the course of action, the elementary movements and the succession of elementary movements. The main following result obtained: the characteristics of expertise are found at the level of elementary gestures and their succession. The latter seem to be defined based on the expression of certain capacities: flexibility and adaptation, anticipation, perception, selection and utilisation of information. The course of action does not characterise an expertise level. The artisans of both group 1 and group 2 are capable of producing stable sequences. On the other hand, both groups produce sequences that vary from one bead to the next, whether with stone or glass. In both cases, the observed course of action can lead to good or bad results. In addition, the knowledge of a method and its variants is not sufficient for the successful production of the desired final product. An artisan can have full knowledge of the strategy that must be followed and still be incapable of transforming a roughout into a sub-ellipsoid preform. In other words, it is possible for an artisan to have practised a method for 20 years, and nonetheless be incapable of fabricating an object of high quality.

These results raise the question of the complexity of the elementary gestures produced by first hominids when knapping stone and their specificity as compared to the technological gestures produced by non human primates. The ancient knapping gestures were followed by a development of various techniques and methods as shown by lithic artifacts. We make here the hypothesis that the knapping gestures reveals cognitivo-motor skills proper to humanoids. These skills enabled the first hominids to discover the possibility to exploit the mechanical properties of the hard stones to create cutting objects. These skills have been actualised given probably a particular socio-cultural context, a point which may be of importance, but difficult to handle.

Characterisation of knapping skills led us to conduct a second field experiment in December 1998. Preliminary results confirm our first hypotheses on the importance of the elementary gestures in the construction of the knapping skills.

Department of Psychology, Tufts University, Medford, MA, USA

Learning by imitation is an extremely efficient mechanism for acquiring complex, object-directed skills. Moreover, imitation permits technological skills to accumulate over generations. Therefore, some scholars have argued that the propensity to imitate is what sets humans apart and makes our tool-using and tool-making abilities unique among animals (cf. Donald, 1991; Meltzoff, 1995). We agree that the ability to imitate is important, but maintain that for tool-using to be as pervasive as it is among humans, something more is involved. Individuals learning through imitation do not limit themselves to acting with just the one object they saw demonstrated or just the one purpose for which they saw it used. Instead they generalize the learned behaviors to other similar objects and to analogous problem situations. This transfer or "extension" of activities learned through imitation is what makes tool-using truly generative.

We have been focussing on this issue of transfer in our research with human infants. By manipulating the means and the ends of goal-directed activities in various ways, we are working to identify what infants "take away from" initial instances of imitation to utilize later in different circumstances. Our developmental findings may be of interest to scientists trying to define the evolution of tool-using and tool-making. Like infants, early hominids presumably did not have access to symbolic instructional media (i.e., language, diagrams, etc.), but they did operate in a rich social milieu. Thus the cognitive problems posed by learning object-directed skills and the potential set of solutions to these challenges are arguably similar for infants and early hominids.

Our initial approach to investigating infants' abilities was to simply look at what infants attended to while they observed goal-directed actions. We showed 15-month-old infants mechanical toys which an adult model operated by pressing, turning, or pulling a handle at one end. The toys were unusually long, so that the frog, for instance, was about a meter away from the handle which made him hop. This feature allowed us to determine whether the child looked at the means, the ends, or both during the demonstration. We found that about half of the infants looked exclusively at the effect during the demonstration. They never looked back towards the experimenter's hand. When the toy was offered to them, these babies tried to produce the effect directly by grasping the animal end of the apparatus. The remaining infants were more active observers. They glanced back and forth between the moving animal and the experimenter's hand during the demonstration. When these babies were offered the toy, they went right for the handle at the far end and manipulated it appropriately. Thus focussing on the means led to successful reproduction of the ends, while focussing on the ends undermined success.

We have also investigated infants' relative attention to the means versus the ends using a transfer paradigm. In this study, 14-16 month-old infants were first shown how to work two novel objects. The demonstration toys were similar to those used before but not so long. They had different handles which worked by different actions to produce different effects. For example, an infant might be shown how to turn a faucet to make a dog jump and how to pull a D-ring to make a bird spin. After the infant had learned to work the demonstration objects, a third object was offered with no demonstration. This test object was a hybrid; it had the handle of one demonstration object and the toy animal of the other. The question of interest was whether infants would try to work the test object by pulling or by turning, that is, according to its handle (the means) or according to its animal (the ends). Most infants acted on the test object as they had before with that kind of handle, suggesting that similarity of the means governed motor transfer for them. However, other analyses showed that infants also tended to employ the motor scheme they had become most proficient with during the learning phase or the scheme they had used with the second (most recent) demonstration object.

We are now further investigating infants' ability to transfer according to the means, using sets of objects having more dramatically different ends. Infants are again taught to work two mechanical toys. This time the two toys have the same handle which operates by the same action, but otherwise the toys are quite different, and the effects the action produces are also distinct. After the infant learns to work these toys, a third toy with the same handle but quite different otherwise from the first two toys is offered without demonstration. Thus far, 8 out of 9 infants have gone right for the test toy's handle and manipulated it with the same action that worked with the demonstration toys. These preliminary results suggest that infants readily transfer knowledge of familiar means to novel objects and expect that some (unknown) effect will ensue.

Finally, we are investigating whether infants can adapt a known means to an object in a novel spatial orientation. An adult model shows infants how to operate a toy laid flat, for instance, and then offers them he toy positioned vertically, so that a different axis of movement is needed to produce the effect. Preliminary results indicate that infants are able to "mentally rotate" the observed means in order to successfully produce the effect; they either appropriately alter the direction of the demonstrated action or they restore the toy to its original orientation and then imitate the action.

Our program of research indicates that preverbal infants exhibit attention, generalization, and flexibility with respect to the means in goal-directed tasks; these abilities permit considerable generativity in their exploitation of objects. This contrasts with non human primates, who may confine their success to single effects because they tend to ignore the means when imitating object-directed actions (Tomasello, et al; 1993). Thus, transfer according to the means might be unique to hominids and could underlie their (our) superior achievements with tools.

Scottish Primate Research Group, University of St Andrews

Tool-use by non-human primates has attracted disproportionate attention, because of a focus on made-objects that is inevitable in archaeology and paleoanthropology. Clearly, however, it is the animals’ manual skills, enabling the objects to be made, that are most important in understanding the evolutionary origins of stone-tool manufacture by hominids. The aim of this talk is to characterize the organization of manual skills shown by chimpanzees and gorillas, whether these are employed in tool-manufacture or not. Much of the evidence will in fact come from observation of apes gathering plants that present physical problems for handling and consumption, and the effects of hand and arm injuries will also be considered. A case will be made that these species, like ourselves, are able to build up novel hierarchical structures involving regular sequences of elementary actions, in modular organization with the option of iterated subroutines, and well co-ordinated manual role differentiation. Further, great apes appear able to use imitation of skilled practitioners as one source of information for this process, implying some ability to ‘see’ behind the surface level of action. This process does not necessarily involve understanding cause-and-effect and the intentions of other individuals.

Departments of Health, Kinesiology, and Leisure Studies and Psychological Sciences, Purdue University, USA

Recent experiments (Savage-Rumbaugh & Lewin, 1994; Toth et al., 1993) have shown that non-human primates can learn to knap stones, however, they cannot do it as effectively as hominids. Possible explanations for these differences in skill proficiency might depend on several species-dependent factors, such as anatomical, biological, or cognitive differences. I this paper, I will suggest that differences in stone knapping proficiency between human and non-human primates might be related to complex interactions between preferred locomotor posture, lateral stability, and the ability to develop fine bimanual coordination. In particular, this paper will argue that efficient stone knapping can only emerge as the result of developing both a stable hand preference and a stable upright posture.

To demonstrate this point, I draw data (1) from my own research on the development of bimanual coordination and laterality in human infants and (2) from recent findings on non-human primates’ laterality. I will present data that will support the following line of arguments:

1) Within their first year of life, pre-locomotor human infants can learn sophisticated bimanual patterns to solve complex tasks demanding the organization of sequenced and complementary movement patterns. Infants can develop and discover how to organize these bimanual patterns much earlier than thought before. This learning is not the result of explicit teaching or demonstration, but rather the product of simple, limited and repeated task exposure (Gadacz & Corbetta, 1999).

2) Early, when infants discover how to use such bimanual patterns, they also use a clear and relatively stable division of labor between hands. This division of labor, however, disappears a few months later when infants begin to develop prone locomotion on their four limbs (Corbetta & Thelen, 1999, in press).

3) Further disappearance of division of labor occurs when infants begin to adopt a bipedal posture and initiate to walk independently. Division of labor, however, seems to reappear a few months later when infants acquire a relatively stable gait pattern (Corbetta & Gadacz, 1999).

4) Further extensions of these data with other human infants studies (Corbetta & Wilson, 2000) and studies on posture and laterality in non-human primates (Spinozzi et al., 1998; Westergaard et al., 1998) seem to confirm the existence of a tight link between locomotor posture and the lateral organization of the upper limb system. Converging results show that crawling or quadrupedal locomotion seem to alter or hinder the expression of stable hand preference and division of labor, while independent walking or the adoption of an upright posture seems to enhance the formation of stable lateral preferences.

In conclusion, I will argue that the development of fine motor control skills such as the one required in stone knapping might depend on the former development of a stable upright posture, which in turn, allows the development of a clear division of labor between a postural and a manipulatory hand. As argued by others (i.e. Corballis, 1991), adopting an upright posture helps free the hands from their postural/locomotor role, and thus, contributes to the development of more sophisticated, better controlled, and more specialized manual activities. This seems to be a uniquely hominid characteristic.

*UMR 5808 du CNRS, Institut du Quaternaire, Avenue des Facultés, 33405, Talence, France

** Palaeo-Anthropology Unit for Research and Exploration, Department of Palaeontology, University of the Witwatersrand, Private Bag 3, Wits, 2050, Johannesburg, South Africa.

Ever since Dart interpreted certain bones from Makapansgat as tools, scientific consensus has fluctuated as to whether bone objects from early hominid sites should be interpreted as artefacts, or the result of non-human taphonomic processes, which are known to produce pseudobone tools morphologically similar to human modified or used artefacts. Here we present possible evidence of bone tool shaping from Swartkrans (Members 1-3; ca 1.8-1 Mya). Four horncores and the proximal end of an ulna used as tools have facets covered by parallel fusiform striations characteristic of grinding which would place them as the oldest known shaped bone tools.Identification of these traces as possibly resulting from deliberate shaping is based on the characterisation of the use wear pattern and other taphonomic modifications observed on the bone tools and study of the remainder of the horncores from Swartkrans. We also used comparative microscopic analysis of naturally weathered horncores, horncores from other Southern African Plio-Plistocene sites (Sterkfontein, Makapansgat, Gondolin), LSA bone tools shaped by grinding, and experimental manufacture of bone and horn. Morphometric analysis of Swartkrans long bone shaft fragments used as tools in comparison with unused bone, indicates that early hominids selected for heavily weathered, longer, wider and more robust pieces. Our results suggest that early hominids had the cognitive ability to modify the functional area of bone implements to achieve optimal efficiency.

Center for the Integrative Study of Animal Behavior, Indiana University, USA

I suggest a cost : benefit : milieu approach to identify variables involved in the acquisition of manual specializations, and to describe ranges of behavioral possibilities. Animals acquire manual specializations when: (1) the manual specialization offers an advantage over the absence of such task specialization, and (2) either (a) the animal is free of conditions that inhibit the acquisition/performance of this behavior or (b) benefits of manual specialization outweigh the costs (temporal, energetic) of overcoming inhibitory conditions. Variables that constrain or limit the acquisition of a manual specialization interact with variables that favorably influence acquisition in complex ways. This approach weighs possible benefits gained by the acquisition of a specialized manual skill (such as a new food source) against costs (such as physical, biomechanical limits of the organism; time/energy expenditure), within the species-typical milieu (external and internal: environmental, social, physiological). Additionally, identifying different types of motor skill learning (chance discovery, imitation, active teaching / learning, trial-and-error problem-solving) addresses the question of how an animal comes to "recognize" the goal of the manual task.

A predictable relationship exists between manual specialization and consistent individual hand preference (not to be confused with species-wide, directional right- or left-handedness). My investigation of variables that constrain and influence the expression of individual hand preferences in five species of wild African monkeys and apes (Holder 1999) illustrates how a cost : benefit : milieu approach can shed new light on old questions by minimizing anthropocentric bias and assumptions.

Finally, I discuss methodological and theoretical problems associated with the interpretation of hominid behavior, whether based upon extrapolation from extant primate behavior (human and non-human) or from interpretation of the hominid (fossil, lithic, environmental) record.

Institute of Archaeology, The Hebrew University, Jerusalem, Israel

In the course of the years 1994, 1999-2000, the Hadar Research Project, conducted by the Institute of Human Origins (ASU, Tempe Arizona), engaged in a systematic survey of the ’upper’ Kada Hadar member above the major disconformity surface (MDS) in the Hadar Formation (Afar, Ethiopia). The survey revealed the existence of archaeological occurrences in the Makiamitalu Basin, where flakes and cores/‘core tools’, associated with faunal remains of both large and small mammals, were found freshly eroding from the slopes of the hills. All the sites are situated in an similar stratigraphic position, within a unit of massive silts, capped at places by the remnants of sandstone. The silt underlies the BKT-3 tuff preliminarily dated to 2.33±0.07 myr (Kimbel et al., 1996). Two archaeological localities - A.L. 666 and A.L. 894 - have been explored to date. In A.L. 666 the remains of Homo sp. (Kimbel et al. 1996, 1997) were found in association with the archaeological material. This contrasts sharply with earlier hominid remains in the region, dated to 3.4 to 2.9 myr, all of which were identified as Australopithecus afarensis (Lockwood et al., 2000, and references therein). Surveys in these earlier members have not yielded any evidence for stone tools in situ.

The lithic assemblages of both sites consist mostly of flakes and flake fragments, ranging from 1 cm to 15 cm in length, with the majority falling in the smaller size fraction. Cores and core tools, while present, are not common. No retouched artifacts were recovered from the excavation at A. L. 666, a feature shared with the ~2.5 myr old assemblages from the Gona drainage on the western margin of the Hadar site (Semaw et al., 1997, 2000) and with the later KBS industry of Koobi Fora (Isaac, 1976). Some flake characteristics (e.g., the amount of cortical cover and its placement on the flakes’ surfaces), as well as those of the few cores and manuports, indicate the use of well rounded cobbles and pebbles. Raw materials at A.L. 666 were identified as basalt and more rarely rhyolite, similar to the composition of cobbles in the stream conglomerates above the MDS.

So far, only the artifacts from A.L. 666 have been subject to refitting attempts and detailed technological studies. The few refits and the technological analysis of this small assemblages (160 artifacts) indicate that rounded cobbles have been used as cores; core morphology likely consisted of a relatively flat surface opposed to a convex one, and was achieved by either splitting intentionally a rounded cobble or by removing fully cortical flakes in the first stage, in order to produce adequate surface geometry. Flakes in the assemblage often present large bulbs of percussion which had left deep negative scars on the core surface, thus suggesting hard hammer and forceful blows employed to detach the flakes. A series of removals were detached from a single striking platform until the angle of the striking platform was no longer adequate for flaking and the surface no longer flat, at which point the core was discarded. Some of the larger flakes in the assemblage bear single scars, whereas striking platforms are usually punctiform , plain or sometimes cortical, indicating lack of platform preparation before removal. Short, simple core reduction sequences are indicated by these data. However, some flakes bear centripetal scar patterns and are devoid of dorsal face cortex, reflecting longer and more elaborate core reduction sequences, possibly with some core re-organization along the line. A relatively large number of flakes are side- struck.

Although not yet studied in detail, similar features appear to characterize the larger assemblage (ca. 860 artifacts collected so far) from site A. L. 894.

If corroborated by further research, these observations suggest that the makers of stone tools in the sites of the ‘upper’ Kada Hadar member were familiar with the basic properties of conchoidal fracture mechanics and possessed the ability for spatial organization of their action, which is necessary for successful sequential flake removal off a core surface (Wynn, 1981). In these fundamental characteristics, the Hadar stone tools are similar to those from Gona (Semaw et al. 1997, 2000) and from site of Lokalalei 2C in West Turkana (Roche et al., 1999; Roche and Delagnes poster abstract). However, stone knapping was conducted differently than seen in the latter case, the main difference being the use of strong blows for detaching flakes, combined with less frequent core re-organization and angle corrections, in turn leading to shorter sequences of reduction. The technological characteristics of the Hadar and Gona sites appear to be similar, whereas the two groups of assemblages exhibit a different technique of core modification and flake production than is seen in the broadly contemporaneous site in West Turkana. The differences may be the result of the differently-shaped packages of raw material used in these two regions (i.e., in the morphologies of the cobbles used for knapping). Alternately, these differences may result from more fundamental processes, such as the occurrence of incipient technological traditions or as yet unspecified changes in the cognitive and/or dextral abilities of hominids over the short time span separating the archaeological occurrences in West Turkana from those in the Afar. Until such questions are answered, the issue of stasis in the earliest known period of stone tool production remains unresolved. On the other hand, the growing database of Plio-Pleistocene lithic assemblages underscores the flexibility in processes of lithic production that characterizes already these early assemblages, suggesting that lithic technology has not followed a gradual evolutionary path.

Institution: Institute of Higher Nervous Activity & Neurophysiology, Russian Academy of Sciences

Butlerov Str. 5a, Moscow 117865 Russia

The Lesgaft Academy of Physical culture, St.Petersburg, Russia

The biomechanical organization of the stroke is the consequence of the anatomy of multijoint system involved in the movement, and of the nervous control of this system. The nervous control is, in turn, defined by the goal of the movement. In the case of stone knapping the multijoint system under control is the human arm, and the goal is to execute powerful and, at the same time, exact stroke. The same multijoint system and the same goals of the movement are characteristics of the tennis stroke. The tennis stroke will serve as an example for illustration the biological mechanisms supplying the regulation of the exactness and of the power of multijoint movement. The requirements of stroke exactness and of stroke power are provided by different mechanisms of movement control. The exactness is defined by the spatial and temporal organization of movement and by the co-ordination of joint torques. The power is defined by the racket mass and speed. To make the stroke efficient the quantity of movement (the power) should be transmitted in the optimal way from the distal to proximal segments and, finally, to the racket. It is the extremely difficult problem from biomechanical point of view. The muscles acting on the distal segment not only provide it with a certain power, but also provoke the forces of interaction between all others segments of kinematic chain. The interaction forces can to make difficult the transmission of the power from distal to proximal segments. Thus, the optimal organization of the stroke consists in the choice of muscles and the periods of their actions, which allows avoiding the disturbances of power transmission. Specifically, the muscle forces should be applied during the semi-period of fundamental oscillations of the segments that assures maximal velocities (and zero accelerations) of all segments at the moment of the stroke. More segments are involved in the movement, more fine the control can be achieved and more powerful the stroke can be. The top-level player uses all degrees of freedom of kinematic chain of the body, i.e. rotations in the leg joints (ankle, knee and hip), in the spine, in the arm joint (shoulder, elbow and wrist). It allows performing the extremely powerful stroke. At the same time the complex control of the co-ordination between the segments is required. The beginner excludes the majority of kinematic degrees of freedom simplifying the control and, as the consequence, imposing limitations on the stroke power. The additional difficulty of motor control is to take into account the unpredictable changes of the environment keeping the stroke exact and powerful. The changes of the internal state (fatigue, trauma) also need the adaptability of motor control. Often this adaptability is related with the high ability to anticipate the different kind of disturbances.

CNRS FRE 2361, 45 rue des Saints Pères, 75270 Paris

Apraxia is a high level motor disorder due to cerebral lesion in which transitive movements like tool use movements are known to be impaired. Apraxic patients may be completely unable to use tool for several reasons, because they can't retrieve the tool's function, or because they don't know the action any more. They may realise the action, but in an approximate way, for example because the posture of the hand for holding the tool is wrong. These observations are well documented in numerous clinical studies. It was also shown that apraxic patients made disordered 3D movements when imitating tool use (pantomime). However, a quantitative analysis of tool use has never been performed in these patients. The aim of this preliminary study is make a preliminary quantitative analysis of tool use, in particular hammering in apraxic patients. To this purpose, we analysed the trajectory and kinematics of the upper limb and the integration of the tool itself in the kinematics of the movement. Six apraxic stroke patients were recruited for this study. Five of them suffered from an extensive cortical lesion of the left hemisphere, including the parietal lobe. The sixth suffered from a lesion in the left fronto-temporal region. All of them were right handed, hemiparetic on the right side, and then performed the movements with their left arm. Six healthy subjects, age matched to the apraxic patients, composed the control group. They all were right handed and performed the movements with their left arm. Movements were recorded using equipment, which allows recording the 3D position and orientation of electromagnetic sensors (Polhemus). Three tool use movements were studied: hammering a nail, turning a spoon in a bowl and slicing bread. These movements were performed with both tool and object. Before movement recording, the tool was placed in the patient’ s hand in the correct orientation. We made two series of recording: the first one with four sensors placed on the left arm, and the second one with one sensor on the left hand and another one on the tool. Pantomime of tool use was also recorded. Turning and slicing movements made by most apraxic patients were similar to those executed by healthy subjects. However, apraxic patients executed the movement more slowly and used more the spatial cues provided by the environment of the task to produce correct movements (for example they followed the wall of the bowl with the spoon). For hammering movement in healthy subjects, motion speed was faster for the head of the hammer than for the hand. This speed amplification resulted from a rotation of the hand and tool complex. Apraxic patients, except one, didn’t show this speed amplification: they just translated hand and tool complex. Thus, they were not able to take advantage of the mechanical properties of the hammer, in order to enhance the efficiency of their action. The only apraxic patient who didn’t show this deficiency in integrating the mechanical properties of the tool in the kinematics of movement was the one with a fronto-temporal lesion. Then, it seems that this deficiency may be specificof parietal lesions. This work allowed us to show i) that apraxic patients were able to produce qualitatively correct movements when spatial cues and constraints were important enough (turning and slicing movements) ii) a deficiency in hammering movements, which may be relevant for further comprehension of the higher levels of integration in motor control. This deficiency seems to be specific of parietal lesions, but further studies are needed to determine whether it’s specific of apraxia.

Mon intervention portera sur trois plans principaux :

Le premier, paradigmatique, définira les conditions théoriques et opératoires des rapprochements possibles entre grands singes actuels et hominidés anciens et plus particulièrement comment poser les problèmes récurrents et généralement non traités du choix des espèces comparées et des télescopages temporels que nous opérons dès lors que nous associons des espaces-temps aussi différents que ceux des hominidés du plio-pléistocène et ceux des chimpanzés d'aujourd'hui. Nous cadrerons ces questions au sein du groupe des Hominoïdes et tenterons d'évaluer ce que peut signifier l'innovation que constitue la taille de la pierre parmi d'autres innovations techniques animales. A une échelle plus large la comparaison interspécifique permet d'estimer l'originalité de cette innovation au regard d'autres innovations d'importance évolutive comparables, j'entends : l'outil utilisé, ("tool-using", extérieur au corps), l'outil fabriqué ("tool-making"), le système technique (plusieurs outils fonctionnant ensemble) ou l'outil secondaire (outil servant à fabriquer un autre outil) que l'on observe à des degrés de sophistication variés dans des groupes taxonomiques très différents, avant même que de poser la question de la taille de roches dures comme élément crucial dans le phénomène d'hominisation.

Le second point de mon intervention portera sur certains des résultats du travail comparatif que je mène sur les choix techniques et l'analyse des actions élémentaires sur la matière anthropoïdes (chez les chimpanzés en particulier). J'ai pu en effet montrer (Joulian 1995) que certains gestes techniques des chimpanzés (le broyage ou la découpe vers soi) étaient absents de leur registre d'activités mais sans que l'on puisse cependant encore montrer quels étaient le ou les facteurs déterminants - biomécaniques, sociaux, techniques, culturels ou cognitifs - en jeu. Cette typologie des formes d'actions possibles et/ou présentes chez les anthropoïdes me permet toutefois de cerner des choix chimpanzés (Joulian 1996, s.p.) et de faire le lien avec les formes d'actions connues pour les hominidés anciens, et d'aborder, dans un troisième temps, la question des connaissances techniques des Hominoïdes.

Dans cette communication, j'attaquerai cette question par un angle particulier, celui des "principes mécaniques d'action" qui sont les représentations des principes mécaniques (pression, traction, torsion, levier, …) mobilisées et déductibles des activités ou des instruments eux-mêmes chez les chimpanzés et les hominidés anciens. Je définirai à quel niveau il convient, me semble-t-il, de les aborder : en bref, ni celui des compétences mentales, ni celui des contenus). La conception de l'activité que je développe est d'inspiration Gibsonienne et se distingue des approches expérimentales ou cognitives de Povinelli (2000) ou de Whiten (1997, 2001) en ce qu'elle prend en compte les propriétés de l'action (Bril, 199) et le poids de l'environnement artefactuel (Joulian 1998, Ingold, 2000) et aborde les principes mécaniques à un niveau intermédiaire (contre une "folk-physics" ou une "folk-psychology" culturellement variables et non pertinentes pour l'échelle d'analyse et de comparaison "Hominoïde" qui est celle à laquelle nous faisons face quand nous nous interrogeons sur la taille de la pierre). J'essayerai de répondre à la question des principes mécaniques d'action techniques chez les chimpanzés d'aujourd'hui, chez les hominidés fabriquant d'outils lithiques et plus généralement chez les Hominoïdes et m'interrogerai 1) sur les capacités cognitives mobilisées dans ces différents principes et 2) les ordonnerai en terme spécifiques (en traitant la question corrélative de leur variabilité inter et intraspécifique).

University of Illinois at Urbana-Champaign, Urbana, USA.

The paper will examine the characteristics of artifact production using ethnographic and historical information and attempt to serve as a background against which to view the tool making of non-human primates. Solving the problem of producing a material artifact draws on a constellation of the practitioner’s resources. Production is a situated activity in which person, means and intent are unified. To an observer, the application of these apparently indivisible resources by the adept, flows in a smooth and uninterrupted fashion toward the intended goal. These are the unreflected motions seen as characteristic of an expert in a given activity. Consideration and analysis seem impossible for the practitioner as unaccustomed attention to any single feature of the process destroys the momentum required for effective accomplishment. Yet the expert is ever sensitive to specific contingencies and possesses the necessary flexibility to modulate actions effectively. Plans must contain the material for emergent restructuring. Paradoxically the very fragmentation which will erode speed, quality and predictability is necessary for the initial integration of these resources while learning an activity and their continued application in a fruitful manner. The contrast between the smooth performance of the expert and the halting efforts of the novice reflects the creation of bonds among the parts of the productive act. Associations of these parts are not immediate but develop and strengthen over time as the activity is repeated. The ongoing action is at once forward moving and reflexive. Its resilience lies in the structural potential of shifting attention from means to end and back again as circumstances warrant. What may appear as a tightly orchestrated unit in fact requires constant monitoring and accommodation to variations which also are part of the dynamic productive process. A challenge for approaching non-human tool making is the validity of deriving conceptual processes such as learning , planning and monitoring from actions. How does one arrive at the unified whole if only one part is accessible?

Tulane University, Department of Psychology, USA

The development of tool use in humans has often been treated as a discontinuous ontogenetic achievement, dependent on children attaining a new level of representational or relational thinking (Bates, 1979; Piaget, 1954). According to this view, young children’s initial attempts to use tools involve a good deal of trial and error behavior, which may or may not result in children using the tool correctly. With the onset of more advanced cognitive skills, however, tool use is said to undergo dramatic changes. Trial and error behaviors disappear as children begin to induce quickly how even unfamiliar tools can be deployed successfully to achieve desired ends.

In this paper, I offer a challenge to this traditional developmental account. I suggest that there is much to be gained from both ontogenetic and phylogenetic perspectives by viewing tool use as a problem of perception-action development (Gibson & Pick, 2000; Lockman, 2000, Smitsman, 1997). More specifically, I argue that tool use is an outgrowth of early perception-action routines that infants have been employing throughout the first year to explore objects and surfaces in their environments. According to this view, tools change the properties of young children’s effector organs. As a consequence, young children need to learn about the new action capabilities afforded by the tool attached to their arms and hands. Trial and error behaviors, rather than signaling a failure of tool use, should be viewed as self-generated opportunities for perceptual learning. More broadly, this view suggests that tool use is a more continuous developmental achievement that emerges from infants’ attempts to explore and gain information from their environments. Phylogenetically, this view also suggests that tool use does not represent an abrupt advance in our abilities or even that of our primate relatives. Instead, the seeds of our remarkable capacity to use and fashion tools may be rooted in infant perception-action routines, particularly those involving the hand and arm.

In support of this view, I will present research that suggests a link between infant object exploration and infant use of tools. The action pattern that I will focus on is banging or pounding objects. I hope to show that infants employ this action pattern not only when they explore objects with their hands but when they begin to use tools like mallets or hammers.

In the developmental literature, infant banging or pounding of objects has often been viewed as a rhythmical stereotypy (Thelen, 1981) that infants employ indiscriminately (Piaget, 1952). In contrast, I suggest that the action patterns that infants evidence in banging may be used instrumentally and incorporated into such tool use behaviors as hammering. In support of the idea that even young infants use banging in a selective and instrumental manner, I first present the results of a longitudinal study on banging (Lockman & Wright, in preparation). Infants were studied monthly from six to ten months of age. Our primary question was whether infants use banging instrumentally and selectively to produce noise with objects. During each testing session, infants were presented with pairs of identical cubes that varied in terms of their hardness. One pair was composed of wood, one pair composed of sponge and one pair was composed half of wood and half of sponge (i.e., the mixed pair). For this latter object, to produce noise, infants had to align the wood side of the cube with the hard table surface if banging the cube singly. If banging the mixed cubes together, infants had to align them so that the two wood sides collided. Our results indicated that even by seven months, infants were using banging selectively. Infants were more likely to bang the hard cubes-singly or together-than they were the soft ones. For the mixed pair, however, infants only evidenced selective banging when banging the cube singly, but not when banging the cubes together. Taken together, these finding suggest that infants use banging in an instrumental fashion to explore surfaces in their environments. But the findings also suggest that aligning objects in relation to one another to produce desired ends may pose special problems for infants. This type of coordination problem may help us understand why some types of tool use tasks are more difficult for children than are others.

Given that infants employ banging instrumentally under some circumstances, we next asked whether infants can use objects that are more tool-like in an instrumental fashion. In a separate longitudinal investigation, we studied infants monthly from eight to ten months and gave them hammer-like objects that were attached to a cube (Cralley, Ellman, & Lockman, 1999). The rigidity of the mallet varied, however. In some instances, the mallet was hard and in other instances the mallet was soft. The results indicated that even by eight months, infants were more likely to bang the hard than the soft mallet, when holding the handle. These findings illustrate some important continuities as well as extensions of the banging study described previously. First, infants are using a handle, not as an object in its own right, but as an extension of an object. This is clearly a critical element of effective tool use. Second, with these hammer-like objects, infants are employing banging in an instrumental or exploratory fashion to effect a change (i.e., noise production) in their environments. More broadly, I suggest that this finding is consistent with the idea that tool use in young children involves the incorporation of actions that are evident in infants’ manual exploratory routines.

In concluding, I will speculate on the phylogenetic implications of these ideas. I will suggest that by considering the ontogenesis of tool use as a more continuous achievement, we may be in a better position to understand how tool use has evolved. Rather than considering tool use as an abrupt ontogenetic as well as phylogenetic advance, tool use may have emerged from capacities or potentials that were already available to the organism. Our evidence on how banging is initially used by infants and how it is incorporated into tool-like situations is consistent with this idea.

1INSERM U.483, Université Pierre et Marie Curie, Paris-6, France

We have attempted to put into relation the phylogenetic development of the direct and indirect corticospinal system and the evolution of dexterity and tool use, as compared between four species: cat, squirrel monkey, macaque monkey, and homo.

A striking variation in dexterity is observed between these four species: in the cat, with an index of dexterity of 2, there is only a very restricted capacity for independent digit movements. Although the squirrel monkey (index = 5) has a much more advanced hand function than the cat, there is no precision grip. The macaque (index = 6) has the capacity for true tip-to-tip opposition of thumb and index in the precision grip, and these functions are further refined in the hand of apes and humans (index = 7). This progression is tightly linked to the degree of the direct cortico-spinal (corticomotoneuronal [CM]) connections to upper arm and hand motoneurones. Thus, it has been concluded (Kuypers 1981, Heffner and Masterton 1983, Porter and Lemon 1993) that the relative development of the CM system is important for the degree of dexterity.

There is, however, considerable debate as to the relative importance of direct CM versus indirect transmission via propriospinal neurons (PN) of corticospinal excitation to cervical motoneurones. We have therefore investigated direct and indirect corticospinal transmission in two species with different index of dexterity: in the New World squirrel monkey (Saimiri sciureus) and in the Old World macaque monkey (macacca fascicularis).

Intracellular recordings were made from motoneurones identified from the ulnar, median and deep radial nerves in adult squirrel and macaque monkeys under chloralose anaesthesia and neuromuscular paralysis. Excitatory and inhibitory post-synaptic potentials (EPSPs and IPSPs) to stimulation of the contralateral medullary pyramid were recorded before and after a lesion to the dorsolateral funiculus at C5: before the lesion in order to assess direct corticomotoneuronal effects, and after the C5 lesion, designed to interrupt direct corticospinal inputs to the lower cervical segments, to unmask indirect (PN mediated) effects.

Before lesion, monosynaptic excitatory effects were found in about 75% of the motoneurones in the macaque as well as in the squirrel monkey. However, the average amplitude of the CM EPSPs was about three times smaller in the squirrel monkey than in the macaque (0.6 mV and 1.9 mV respectively). The cat, in comparison, has no CM connections.

To estimate the influence of indirect (PN) effects a lesion was made at C5. This lesion, as expected, greatly reduced the proportion of motoneurones showing either direct CM EPSPs or disynaptic IPSPs. In the squirrel monkey, the proportion showing indirect, most likely PN-mediated EPSPs with segmental latencies beyond the monosynaptic range, was large and unaffected: 86% of the motoneurones still showed indirect effects compared to 79% before lesion. These results were in striking contrast to the macaque: after lesion, late EPSPs were found in only 19% of motoneurones, compared to 18% before it. In the cat, in contrast, 100% of the upper limb motoneurones receive indirect PN-mediated excitation (Alstermark and Lundberg 1992).

These findings suggest that the positive correlation across species between more advanced dexterity and the strength of the CM system is, in addition, accompanied by a negative correlation between this function and the strength of the indirect PN system. The results indicate that in primates with more advanced hand function, the CM system may have replaced PN-mediated control, and predicts that in man, where the CM system is more highly developed than in any other species, the C3-C4 PN system, as described in the cat, is unlikely to be responsible for significant transmission of cortical commands to upper limb motoneurones.

How is the development of dexterity and its neural substrate related to object manipulation and use of feeding tools in non-human primates? There is a general trend between the degree of development of the CM system and the degree of tool use: species with and index of 5 or 6 use tools but do not manufacture them, whereas apes with a index of 7 do use and manufacture feeding tools. However, a dexterity index between 5-7 does not allow any prediction on the prevalence of tool use or tool manufacturing: among species with an index of 5 and a CM system, such as saimiri and cebus, the latter does show tool use but not the former. Among the apes with and index of 7, pan and pongo do show tool use and manufacture but not gorilla (van Schaik et al 1998). Thus, the development of the CM system in primates seems more strongly related to dexterity than to tool use. This suggests that the existence of direct corticomotoneuronal connections is a necessary and sufficient precondition for dexterity, but is not a sufficient condition for tool use and tool manufacturing.

*Anthropology, Miami University, Oxford, Ohio, USA

**Anthropology & Zoology, Miami University, Oxford, Ohio, USA

Hominid knapping of stone in prehistory likely did not spring full-blown, but what could have preceded it? Among living great apes, there is no natural flaking of stone, but there is elementary lithic technology. Wild chimpanzees (Pan troglodytes) across Africa use anvils and those in far western Africa use hammer-and-anvils in percussive food processing.

Wild chimpanzees from Senegal to Tanzania use anvils of stone or wood to smash open hard-shelled fruits, e.g. Strychnos spp. Wild chimpanzees from Ivory Coast westwards use hammers and anvils of stone or wood to crack open nuts, e.g. Elaeis guineensis, Panda oleosa, Coula edulis, etc. (Conversely, chimpanzees elsewhere in Africa do not crack nuts, even when these species and the appropriate raw materials are available. This behavioral variation suggests cultural diversity.)

We report new data from Mt. Assirik, Senegal, on the use of percussive technology to process the fruit of the baobab tree (Adansonia digitata). For chimpanzees in this savanna habitat, the pericarp and seeds of baobab are a keystone food. The methods were archaeological, as the field season in the spring of 2000 followed the season of consumption, so that only food remnants and stones were available. We "excavated" three baobab processing sites, where we counted, weighed and measured hundreds of stones and fruits, and noted their spatial relations. Using these indirect measures, we compared two hypotheses: That hammers and anvils were used, or, that only anvils were used.

Data were analyzed in terms of the relative location of smashed open fruits to stones of various sizes and conditions (e.g. portable vs. embedded). Fruits processed without percussive technology, by baboons, served as controls. Results indicate that processing was by anvil alone, rather than by hammer-and-anvil.

We suggest that percussive technology originated in ancestral hominoids with hard-shelled fruits being smashed against living tree trunks and boughs. With greater terrestriality, anvils of stone and roots became available. Nuts are too small for hand-held percussion, so a cognitive leap was needed to progress from object-strikes-passive-tool to active-tool-strikes-object-on-passive-tool. For hard nuts (e.g. Panda oleosa) it is necessary that both hammer and anvil be of stone. This creates the opportunity for inadvertent flaking from mishits, when stone strikes stone. The stage was set for the second major cognitive leap, that of knapping to produce a cutting edge.

The precise motor skills needed for nut-cracking (as opposed to fruit-smashing) seem indistinguishable from those needed for knapping. Similarly, experimental studies of captive apes show that they have the cognitive capacity to learn to fracture stone. So, why do wild apes not knap stone? Perhaps for the same reason that bonobos (Pan paniscus), unlike chimpanzees, do not use tools in subsistence. They have no such need.

Research supported by L.S.B. Leakey Foundation and Phillip and Elaina Hampton Fund (Miami University).

Department of Anthropology, Arizona State University, USA

The question of who made stone tools is raised at many sites by the existence of two or more early hominid species at the same level as stone tools, as long ago as the earliest known tools in Africa at 2.6Myr BP. It is of interest to determine whether all the species were making tools, or whether there were perhaps variations among them in the kinds of tools used and/or manufactured. The most direct potential source of evidence regarding this question is the hand bones of these fossil hominid species. The bone shapes and internal structure provide clues to the magnitude and kinds of stresses associated with habitual uses of the hands. Joint surface topography reflects potential joint ranges of motion and loading capabilities. Muscle attachment areas may in some cases serve as guides to the relative sizes of muscles, and skeletal contours in the region of tendon attachments may indicate relative lengths of tendon moment arms. Functional interpretation of these skeletal clues to hand functions and potential behaviors requires detailed knowledge of hand functions (e.g. grips, joint movements and muscle recruitment) associated with habitual manipulative behaviors of living humans and nonhuman species, as well as an understanding of the morphological features that facilitate these functions and accommodate the internal and external loads associated with the behaviors.

We have applied an interrelated set of approaches to obtaining this knowledge and to discerning links between specific behaviors, hand functions and hand morphology. These approaches include (1) analysis of hand grips, movements and muscle recruitment associated with the use and manufacture of prehistoric tools, using electromyography (EMG), videotapes, a CyberGlove and a strain gauge, (2) observation of hand grips and movements used in feeding and tool activities by captive and wild nonhuman species, (3) kinematic and biomechanical analysis of cadaver specimens from human and nonhuman species, comparing joint ranges of motion, tendon excursion, hand muscle cross sectional areas, tendon moment arms and muscle torque potential, and (4) comparative 3D stereophotogrammetric analysis of curvature and area of mutual joint surfaces.

Results of the comparative behavior studies reveal clear differences between human stone tool making and nonhuman feeding and tool using in the functional capabilities they require and in the stresses they impose upon the hand. The ability to tightly hold a stone in each hand by the thumb and finger pads, allowing the working edges to be exposed and resisting displacement of the stones when they are struck together, is essential to the effective removal of flakes from a stone core by a hammerstone. The only human grips that consistently accomplish this activity are forceful precision grips, such as the 3-jaw chuck and cradle grips, which have not been observed in nonhuman species for any of their manipulative activities. In fact, the nonhuman species require two hands to resist large external forces (such as the pull of teeth on fruits) in maintaining grips.

EMG experiments show particularly strong, repeated recruitment of the intrinsic muscles of the thumb, which probably is an important factor in the maintenance of these human forceful precision grips. Evidence from our cadaver studies of significantly larger tendon moment arms and potential torque for most of these intrinsic thumb muscles in humans, compared with chimpanzees, is consistent with this demand of stone tool making. The large moment arms allow more torque for a given muscle size, reducing the amount of energy required for habitual activities requiring strong, repeated and prolonged contraction of the muscles.

Humans also are distinctive in the ability to accommodate the hand to varying shapes of objects. A unique pattern of thumb and finger joint topographical features (examined with stereophotogrammetry) facilitates this cupping of the hand in humans. The pattern includes significantly less curvature of the mutual trapezial and first metacarpal joint surfaces at the base of the thumb compared with chimpanzees (p < 0.001), allowing the thumb to oppose all five fingers.

The fifth finger has emerged from the experiments as an important contributor to the security of human grips. The intrinsic muscles of this finger are strongly recruited when the finger resists displacement of the stone core during flake removal, buttresses the other fingers holding the hammerstone, and anchors the grip of cylindrical tools for pounding and clubbing. Consistent with these functions are the distinctively robust fifth metacarpal of humans and metacarpal joint configurations facilitating supination with flexion, which is essential for accommodating the hand to tool shapes.

We suspect that the differences between humans and chimpanzees in thumb, index and fifth finger joint rotational capabilities and in moment arms of the thumb intrinsic muscles may explain, to some extent, the limited success of chimpanzee subjects in producing flakes by a two-handed percussion technique.

The results of these behavioral and morphological studies are mutually consistent and provide a new framework for functional and behavioral interpretation of fossil hominid hand bones. For example, stereophotogrammetric analysis reveals that curvature of the joint at the base of the Australopithecus afarensis thumb is virtually identical to the mean curvature in chimpanzees, whose curvature is significantly greater than in modern humans, limiting excursion of the thumb toward the fourth and fifth fingers. Complementing this evidence is the lack of robusticity and supination capabilities of the fifth metacarpal. However, the index finger in this species has the full set of human features that would have facilitated an effective 3-jaw chuck grip of small spherical stones for throwing and pounding.

In contrast, the thumb metacarpal surface of the trapezium is almost flat in the Homo habilis hand from Olduvai (O. H. 7), which is contemporary with early stone tools. This flat joint would have allowed extensive excursion of the thumb toward the fifth finger for the grip of large stones, but might have been unstable for firm grips of small objects between the thumb and index finger. The interesting question now will be to discern morphological changes that accompany subsequent changes in the record of stone tool manufacture.

PRI, Kyoto University, Japan

(to be completed)

UMR 7055 du CNRS, Préhistoire et Technologie, France

1) Split breaking and conchoïdal fracture

The cracking of nuts demonstrated by chimps has been frequently compared to the removal of flakes from a piece of hard stone. However, there are important differences between these two percussion movements or techniques. Cracking a nut is in fact very similar to the splitting of a pebble or cobble : the nut or cobble is placed or held on an anvil and a straight impact is delivered on the top of it with a heavy hammer (another cobble or a piece of wood for the cracking of a nut).

Conchoïdal fracture implies a localized impact -that is a very precise percussion-, and an oblique angle towards the platform -that is the control of the direction of the stroke, which is delivered with a lighter hammer acted with much more speed.

The splitting of round pebbles was practiced by some early hominids when small quartz pebbles were the only available raw material. At Lokalelei, however (see abstract by H. Roche and A. Delagnes), precise conchoïdal fracture was used for the systematic production of flakes, demonstrating a true bimanuality (the minor hand orienting the core in the 3 dimensions of space) and a remarkable precision of the successive strokes (there is no crushing of the platform).

According to J. Tixier (1967), the term "technique" should be restricted to the modes of execution of stone knapping (percussion with or without an anvil, shape and material of the tool(s) used, way of holding the piece and body position). "Method" refers to the organization in space and time of the different removals ("reduction process").

Methods of knapping can be recognized by the "technological reading" of the archaeological material, that is by reading the order and the direction of the removal negatives (flake scars) observable on the flakes, cores and shaped tools.

Some methods seem to be reducible to the repetition of elementary removals following a simple formula (algorithm) : for instance alternating removals along the edge of a relatively flat blank, or successive adjacent flakes around the edge of a large flat surface acting as a platform. In such cases, archaeological collections and experiments show that there is no need to monitor the shape of the core, even though the exhausted core will end with a particular shape.

Other more elaborated methods of knapping, the most famous of them being the Levallois method, imply precise control of the shape of the core, where a productive face is opposed to a peripheral platform. In such cases, some "preparation" flakes are detached in order to correct the shape of the core, while others are the "true" products, the shape of which are geometrically predetermined by the preparation flakes. In opposition to simple algorithmic methods, the knapper conducts the process so as to attain particular core shapes (morphology of the flaking surface, morphology of the platform). The succession of removals is thus highly variable, since the resulting flakes can correspond not only to intended products, but must also be precisely organized in order to control the modification of the core. Such an adaptation of the successive flakes towards a planned objective accounts for a specific skill in the mental construction and selection of short and long term sequences of removals ("constructive know-how").

Anthropologists have long been interested in the potential contribution of prehistoric lithic artifacts to understanding the physical and mental capacities of fossil humans. Consequently, stone tool "technologists" attempt to provide an evaluation of the psychomotor abilities implied in the different lithic production systems of the major periods of prehistory. Three major approaches have been applied in this effort:

1 - the application of developmental models of child intelligence, such as that proposed by Wynn, which supposes that phylogenetic development (during the course of hominization) followed the same stages that observed for modern children (ontological development, in particular the model of Piaget);

2 - collaboration between archaeologists and psychologists specializing in the study of motor and psychomotor activities in order to analyze certain modern stone knappers, such as the chalcedony bead makers of Cambay, India (V. Roux, B. Bril and colleagues);

3 - propositions by archaeologists who are skilled at flint knapping, based on auto-observation and auto-analysis of their knapping behavior. Some have had the advantage of witnessing or participating in the process of apprenticeship of beginning students, which can also constitute an important source of information.

Each of these approaches can be subject to criticism.

The first is based on a hypothesis that remains to be demonstrated: the analogy between phylogenetic and ontological developmental stages. Furthermore, it is impossible to avoid the difficulty of attributing a given knapping realization to a given stage of competence (eg, which of Piaget’s "stages" of intelligence is necessary to knap a good quality biface?).

The second research procedure appears to be more solid, but is in fact limited to only part of the phenomenon in question. Even when the most rigorous methods are employed, motor psychologists record only the visible part of a knapping activity (frequency and acceleration of percussion gestures, regularity of the objects produced, sequencing and duration of the principal operations, etc.). The processes that occur in the intimate space of the "black box" -the brain of the actor- are only indirectly perceptible to psychologists. The risk of this type of analysis is thus to neglect the most mental aspect of knapping activities, which if it indeed exists, remains to be objectified.

The third approach is also problematic. As it is based on the non-objective and non-quantifiable "introspection" of contemporary humans, who are more or less educated in the art of "theorizing", this approach is obviously lacking in terms of scientific rigor and validity. Consequently, to our knowledge, the results of such studies have been published by only a few, isolated individuals, even though modern stone knappers often discuss relevant issues among themselves.

For these reasons, a questionnaire has been prepared and distributed among modern knappers, including both amateurs and professional archaeologists. While it is certain that their opinions and introspections will remain "intuitive" (the only quantifiable aspect being the amount of apprenticeship required for each individual to gain confidence in a given type of knapping, which can permit us to rank them in terms of "difficulty"), they might gain in validity if we reveal a consensus among knappers on certain issues.

UMR 7055 du CNRS, Préhistoire et Technologie, France

The workshop focuses on what characterizes complex actions such as stone knapping, which appears among early hominids at least 2.6 millions years ago in East Africa. The aim of this contribution, together with Jacques Pelegrin’s paper, is to show why and how stone tools are the best if not the only elements, which can testify to the technical performance of our ancestors through prehistoric times.

While J.Pelegrin will address the role of physical properties and responses of hard rocks when prompted for adequate breaking, as well as technical abilities and cognitive capacities required for controlled knapping, I will try and demonstrate the specificity of plio-pleistocene lithic productions through archaeological examples, among which the remarkably well preserved 2.3 myr assemblage of Lokalalei 2C site (West Turkana, Kenya). Through the analysis of LA2C assemblage (see also H.Roche and A.Delagnes poster abstract), we can demonstrate relative planning and accurate technical abilities, which lead to large production of cutting flakes (more than 50 pieces for one block).

But if LA2C material shows that groups of late pliocene hominids had advanced mastery of stone knapping, it also contrasts with other less systematized earlier, contemporaneus or later plio-pleistocene productions -which simplicity, however, should not mask the controlled character of the technique employed. This fact should not be set aside when we try to fit early hominids technical performances in evolutionary processes.

Finally we will see that several hundred of thousands years will be necessary before more elaborated stone knapping activities develop. They result, among other technical features, in mass production of normalized shaped artefacts, which standardization (handaxes) and/or predetermination (cleavers) is independent from the morphology of the raw material employed. Chaining of elementary technical gestures then clearly depend on action planning.

UMR 7055 du CNRS, Préhistoire et Technologie, France

It is normally considered that stone knapping developed following a simple-to-complex pattern. If this is true in terms of evolutive trends, new archaeological data moderate this evidence. Among them is the demonstration of reasoned and controlled stone knapping at 2.3 myr by a group of hominids on the west side of lake Turkana (northern Kenya) as shown at Lokalalei 2C (one of the archaeological sites excavated by the West Turkana Archaeological Project in the Nachukui Formation). The most complete refitted sets and the technological analysis of the whole assemblage (2500 lithic elements) show that:

• blocks of similar volumetric morphologies have been used (selected?) as cores, either starting from a naturally shaped block, or from an intentionally broken block, or even from a flake;
• these morphologies always present a plane surface opposed to a convex one, and an adequate flaking angle which most of the time occupies the entire circumference of the block;
• the plane surface is flaked by series of removals organized in such a way that the surface is maintained flat; as a consequence, the abandoned core retains the original morphology of the block; it also can be extremely reduced;
• this systematic debitage sequence results in a large production of flakes (up to 50 for a single block), so far unequalled for a site of this age; a good percentage of these flakes have similar morphometrical patterns;
• among other technological features linked to manual dexterity of the knapper, clear evidences of platform rectification are also observed, that should be related to minimal anticipation within the chaining of gestures.

Lokalalei 2C can thus be considered as a demonstration of 2.3 myr hominids minimal planning and excellent mastery of elementary knapping gestures. But it should also be stressed that similar skills have not yet been observed in other Plio-Pleistocene sites.