Published in Behavioral and Brain Sciences
Volume 25, Number 3: 389-402 (June 2002)

© 2002 Cambridge University Press

Archaeology and cognitive evolution

Long abstract - 197
Short abstract - 118
Text - 12,348
References - 1,732
Entire text - 14,434

Thomas Wynn
Department of Anthropology
University of Colorado
Colorado Springs
twynn@uccs.edu

Archaeology can provide two bodies of information relevant to the understanding of the evolution of human cognition - the timing of developments, and the evolutionary context of these developments. The challenge is methodological. Archaeology must document attributes that have direct implications for underlying cognitive mechanisms. One example of such a cognitive archaeology is that for spatial cognition. The archaeological record documents an evolutionary sequence that begins with ape-equivalent spatial abilities 2.5 million years ago and ends with the appearance of modern abilities in the still remote past of 400,000 years ago. The timing of these developments reveals two major episodes in the evolution in spatial ability, one 1.5 million years ago and the other one million years later. The two episodes of development in spatial cognition had very different evolutionary contexts. The first was associated with the shift to an open country adaptive niche that occurred early in the time range of Homo erectus. The second was associated with no clear adaptive shift, though it does appear to have coincided with the invasion of more hostile environments and the appearance of systematic hunting of large mammals. Neither, however, occurred in a context of modern hunting and gathering.

Archaeology can provide two bodies of information relevant to the understanding of the evolution of human cognition - the timing of developments, and the evolutionary context of these developments. To do this, archaeology must document attributes that have direct implications for underlying cognitive mechanisms. One example of such a cognitive archaeology is that for spatial cognition. The archaeological record documents an evolutionary sequence that begins with ape-equivalent spatial abilities 2.5 million years ago and ends with the appearance of modern abilities in the still remote past of 400,000 years ago. The timing of these developments reveals two major episodes in the evolution in spatial ability, one 1.5 million years ago and the other one million years later.

Thomas Wynn

Department of Anthropology

University of Colorado in Colorado Springs

twynn@uccs.edu

Keywords: Archaeology, symmetry, spatial cognition, evolution, Homo erectus

I have two goals in this article. The first is to make a case for the relevance of archaeological contributions to studies of the evolution of cognition. The second is to provide an example of one such contribution, a reconstruction of aspects of early hominid spatial cognition based on an analysis of artifactual symmetries.

Assuming that human evolution is relevant to understanding the human condition, an intellectual position that is at the core of biological approaches to behavior, if not yet psychology, then archaeology can supply two important bodies of evidence: 1) actual timing of developments, and 2) the evolutionary context of these developments. The challenge is not epistemological; archaeology can and does supply these things to the study of human evolution in general. The challenge is methodological. How can archaeology inform us about the evolution of mind?

Archaeology is a set of methods for reconstructing past action from traces that exist in the present. These traces include objects made or modified by people in the past - tools, houses, ornaments, and so on - but also less tidy patterns like garbage and refuse of all kinds and evidence of past landscapes (through analysis of soils, pollen, faunal remains, and so forth). Because some traces survive the ravages of time better than others the archaeological record is a biased and non-random sample of past action. Stone tools, for example, survive well but wooden tools do not. Also, some environments preserve traces better than others. Tropical environments are poor preservers, but cold, dry, arctic environments are good preservers. Archaeology is an observational discipline. Unlike laboratory scientists, archaeologists cannot duplicate events and unlike ethologists archaeologists cannot depend on obtaining corroborating observations, though we certainly hope for them. There is a real element of serendipity in archaeology. Pompeii, the "Ice man", and the recent discovery of 400,000 year old spears at Schoeningen (Thieme 1997), are unique and wonderfully informative but they are atypical. The archaeological record boasts relatively few such treasures. Instead it consists largely of more incomplete and mundane traces that allow archaeologists to reconstruct some of what occurred in the past. The primary methodological task of the archaeologist is this reconstruction -- translating traces into actions -- and archaeology has developed a large body of concepts and techniques for doing this. We are very good at reconstructing diet from garbage, and social/political systems from the size, character, and location of settlements. Can there be an archaeology of cognition? This is in reality a two-part question. First, can traces of action inform us reliably about any aspect of cognition, and second, if so, can archaeologists overcome some rather serious methodological roadblocks inherent to the archaeological record of such traces?

One way that psychologists learn about the mind is by observing the actions of individuals in controlled laboratory settings or in natural situations. Sometimes these actions leave tangible traces that become the focus of the analysis. Children’s drawing is one example; block shuffling tests are another. The methodological task of the psychologist is to translate the tangible results into meaningful characterizations of the mind that produced them. Of course, psychologists can also talk to their subjects, but in principle psychology can and does analyze the traces of action. An archaeologist trying to do the same faces some additional hurdles. To make a convincing argument in cognitive archaeology, one must be able to identify specific features of the archaeological record that can inform about cognition in a valid and reliable way. This is the crux of the matter. Unfortunately, the disciplines of archaeology and psychology have never shared much in the way of theory and methodology. For an archaeologist to make a compelling case, he or she must not simply refer to a few selected psychological results. There must also be some understanding of the theoretical and methodological context of the research. With this in hand, the archaeologist can define a set of attributes that can be applied to the archaeological record. This definitional step is indispensable. It is very unlikely that variables taken directly from the psychological literature could be applied to archaeological remains. On the other hand, the traditional categories of archaeology are inappropriate, a point that bears emphasis. Over the last century and a half archaeology has developed a large set of categories for the description of archaeological remains. Some of these categories are based on presumed function (e.g., ground stone axe, or temple complex), some on presumed usefulness in temporal ordering (e.g., Iron Age), some on social complexity (e.g., "Classic" Mesoamerica), and so on. None, to my knowledge, has ever been defined with cognition in mind, and it would be misleading to use them as such (e.g., to argue that Iron Age people were cognitively different from Stone Age people). The cognitive archaeologist must avoid using these traditional categories and approach the archaeological record from the perspective of psychological theories and methods.

Even after careful definition, archaeology faces a number of roadblocks peculiar to its data. The first is preservation. Not only does preservation produce a biased record, it also presents a sliding scale of resolution. The farther back in time we look, the worse the record. There is less preserved, and there are fewer examples of what is preserved. This alone gives the archaeological record a misleadingly progressive appearance; 10,000 year-old remains appear more complex than 500,000 year-old remains partly (but not entirely) because we have so many more of them. The second caveat is logical. How can we be sure that archaeological remains are a reliable reflection of the cognitive abilities of past people? Might not these people have invested their cleverest thinking in domains that are archaeologically invisible? There is no infallible way around this problem. Archaeologists can only assess the minimum competence necessary to produce a particular pattern. Our only comfort comes from increasing the number and, especially, the variety of corroborating cases.

Finally, cognitive archaeology works best on an evolutionary scale of resolution. The ultimate achievement of cognitive archaeology would be to provide descriptions of the cognitive life-world of human antecedents at many points in evolution. Such descriptions would provide an evolutionary foundation for understanding the modern mind. I have long harbored the desire to provide a comprehensive account of the mind of Homo erectus, a very successful ancestor who was the immediate precursor of Homo sapiens. Surely, an understanding of Homo erectus’ cognition would illuminate aspects of the modern mind; there must be much of Homo erectus with us still. Unfortunately, I do not think such a comprehensive description is possible; the archaeological record is too incomplete.

Archaeology can take another approach to the question of evolution, an approach not focused on descriptions of individual antecedents, but one focused on long term patterns of change. Even though poor in detail, the archaeological record is very long, providing a quasi-continuous record of products of action that spans over two million years. Archaeologists can use this record to identify patterns of cognitive evolution that provide insights into questions of modern cognitive science. What follows is an example of this approach. The focus is on spatial cognition (generally considered, including shape recognition and image manipulation). The evidence will consist of artifactual symmetry.

2. The Archaeological Record of Artifactual Symmetry

I have chosen to survey the evolution of artifactual symmetry three reasons. First, symmetry is a pattern and a concept that is recognized by everyone, which reduces the requirement for definition (but does not eliminate it entirely). Second, symmetry has been incorporated into many schemes of spatial cognitive development, and also into theories of perception, so that it provides a direct way to articulate the archaeological record with the cognitive science literature. Finally, it is amenable to visual presentation.

There are several different patterns to which we apply the term symmetry. The most familiar is reflectional symmetry, also known as bilateral or mirror symmetry. Here one half of a pattern is duplicated and reversed on the opposite side. Implicit in the pattern is a central line, usually vertical, that divides the pattern into reflected versions of one another. Bilateral symmetry is "natural" in the sense that we see this pattern in the natural world of plants and animals. A second symmetry is radial symmetry, in which a pattern repeats not across a line, but continuously around a point. Similar to radial symmetry is rotational symmetry, in which a pattern is not reflected across a line, but rotated around a point; here the pattern is not reversed or inverted. Finally, there is translational symmetry, in which a pattern is repeated in a sequence, without reversal or rotation.

Symmetry is ubiquitous in the natural world and the cultural world. It is a well-known feature of crystal growth, resulting from the chemical structures of the molecules themselves. It also acts as a principle in biological growth and development, as in the symmetrical duplications of supernumerary appendages in beetles (Bateson 1972), where the source probably lies in the genes regulating growth. On a larger scale, symmetry is a feature of the overall body plans of many organisms, from microscopic foraminifera to large vertebrates. In human material culture, symmetry appears in the form of artifacts, buildings, and built environments all over the world. It is a central component of decorative systems in almost all human culture, and also a component of games (e.g., string games) and mathematical puzzles (e.g., tessellations). In many of these cases the symmetry results from the application of transformational rules; simple figures repeated and "moved" to produce intricate patterns. Symmetry is so fundamental in western culture, at least, that it is often a metaphor for balance and regularity (e.g., the "symmetrical" arrangement of keys in The Marriage of Figaro). Moreover, it is often endowed with meaning, carrying explicit and implicit information about fundamental values of a culture (Washburn and Crowe 1988).

Not surprisingly, perception of symmetry has been the focus of psychological research for over a century (Wagemans 1996). It is now generally accepted, for example, that reflectional symmetry is perceptually more salient than translation and rotation. Indeed, some experimental work suggests that reflectional symmetry can be detected pre-attentively. Reflectional symmetry across a vertical axis is more salient than reflection across a horizontal axis, with oblique orientations falling a distant third. In addition to such empirical generalizations, there are competing theories of symmetry perception, and it remains a component of general theories of perception (Tyler 1996). It has even come to be a focus in evolutionary psychology, where detection of asymmetry is seen to be a means of mate assessment (Gangestad 1997). Given the ubiquity of reflectional symmetry in the natural world, and its correlation with successful ontogenetic development in many, many, organisms, it is not at all surprising that perceptual systems should have evolved a sensitivity to symmetry. It is quite likely, then, that the perceptual saliency of symmetry is not a derived feature of human perception, but is one we share with many complex organisms. The degree to which it is shared, and whether it has evolved independently in several taxa or is instead a very old feature, are interesting questions, but tangential to the current discussion. The archaeological record does not document the development of symmetry perception per se. Instead, it documents the imposition of symmetry on material objects. Detecting symmetry is not sufficient for this task; other cognitive mechanisms must come into play. The importance of the archaeological record of symmetry lies not in the symmetry itself, but in what it reveals about these other mechanisms.

Most of the following analysis will focus on stone tools. They are far and away the most abundant material evidence archaeologists possess for the majority of human evolution. The record of stone tools begins 2.5 million years ago and extends to the present. From the tools themselves archaeologists can reconstruct a variety of actions: raw material selectivity and procurement, manufacturing sequences, use, and discard. Archaeologists have been most interested in reconstructing the specific uses of stone tools, and the role these tools played in subsistence and, sometimes, social life. But these reconstructed actions, those of manufacturing in particular, can also document particular cognitive abilities.

Fracturing a stone produces sharp edges; this is the basic principle underlying almost all stone tools. Archaeologists use the term "knapping" to refer to the stone fracturing process. In the simplest case a stone knapper uses one stone, termed a hammer, to strike another. If the knapper has struck with enough force, and delivered the blow to an appropriate spot at the appropriate angle, the receiving stone, termed a core, will break. In most instances the knapper must direct the blow toward the edge of the core because a blow landing toward the center is unlikely to deliver enough force to produce a fracture. This simple act of knapping produces two potentially useful products, a smaller sharp edged piece termed a flake, and the larger core, which now also may have sharp edges. Even this simplest of knapping actions requires directed blows. Randomly bashing two rocks together can produce useful flakes but even the earliest stone tools, 2.5 million years old, resulted from directed blows. The subsequent development of knapping technology included increases in the number of blows delivered to a single core, greater specificity in the location of blows, modification of flakes, longer sequences of action between the initial blows and the final blows, a greater variety of hammering techniques, and more regularly shaped final products.

[FIG 1]

Recently Stout et al. (Stout, Toth et al. 2000) have conducted a pilot PET study of basic stone knapping using an experienced knapper (Nick Toth) as the subject. The result showed highly significant activation in several brain regions. Much of this activation is what one would expect from performance of a complex motor task based on hand-eye coordination (primary motor and somatososensory areas, and cerebellar activity [p. 218]), but Stout et al. also recognize a significant "cognitive" component, implied by the activation of superior parietal lobes.

In other words, simple stone knapping is a "complex sensorimotor task with a spatial-cognitive component" (p. 1221). These results, though preliminary, situate most of the significant cognitive activation within the "dorsal pathway" of visual processing (Ungerleider 1995; Kosslyn 1999). The ventral pathway associated with object identification and shape recognition is minimally activated, implying that shape is not a significant component of the basic flaking task. These results are preliminary and need confirmation. There was only one subject, a skilled knapper, and the knapping task was brief and basic -- removing flakes from a nodule (a Mode 1 procedure; see below). Nevertheless, it reinforces work of cognitive archaeologists who have focused on spatial concepts borrowed from developmental psychology ((Wynn 1979; Wynn 1981; Wynn 1985; Wynn 1989; Robson Brown 1993)).

The directed action of stone knapping preserves something of the cognition of the knapper. Even in the simplest example, the knapper must make a decision about where to place a blow and how much force to use. These decisions are preserved in the products themselves. It is now common for archaeologists to "refit" cores by placing the flakes back together in a kind of 3-D jigsaw puzzle. Such a reconstruction permits archaeologists to describe in detail long sequences of action including specific location of blows, reorientation of the core by the knapper, and subsequent modification of flakes (Schlanger 1996). But even simple tools can be informative. The pattern of "negative scars" on cores or modified flakes preserves the sequences of blows. Archaeologists interested in cognition can use these preserved action sequences to investigate a variety of cognitive abilities, including sequencing, biomechanical skill, and spatial cognition. Even the simplest knapping required some notions of spatial relations and as stone tools became more complex they often preserved more complex understandings of spatial relationships.

There is a problem with intentionality. All stone tools have a shape, and this shape preserves spatial relationships, but how intentional were they? Here I do not mean the layers of intentionality invoked in theory of mind literature, but the simple question of whether or not a stone knapper intended to produce a particular shape. The basic action of stone knapping will produce useful results without the knapper intending the final core and flakes to have any specific appearance whatsoever. It is even possible for the iterative application of a specific flaking procedure to produce a final core with a regular shape, completely unintended. The shape itself, and the location and extent of modification producing the shape can often, but not always, document intention. For example, the artifact in FIG 5 has extensive trimming on one side that produces a "shoulder" mirroring a natural configuration on the opposite side. This is unlikely to have been an accident.

It is appropriate and traditional to begin discussions of human evolution with a discussion of modern apes. Much of our anatomy and behavior is shared with apes, including characteristics of the brain and cognition. A necessary first step in any evolutionary analysis is the identification of what is peculiarly human, for this allows correct focus of the undertaking. If modern apes, especially chimpanzees, employed all of the spatial abilities used by humans, then our evolutionary understanding must focus on the evolution of apes in general. It is also an axiom of paleoanthropology that human anatomy and behavior evolved out of those of an African ape, so that a description of this ancestor is a logical starting point for any summary. Our best information concerning this common ancestor comes from the living African apes (who, of course, have also evolved, but because their anatomy and habits appear more like those of a "general ape" than the anatomy and habits of the obviously unusual humans, they are a better candidate than ourselves).

Whatever the cognitive requirements of stone knapping are, they are within the abilities of apes, at least at the basic level of using a hammer to remove a flake. Nick Toth and Sue Savage-Rumbaugh have taught a bonobo to flake stone, and the results of their research help identify what might have been different about the cognition of the earliest stone knappers(Toth, Schick et al. 1993; Schick, Toth et al. 1999). Kanzi, a bonobo also known for his ability to understand spoken English and use signs, learned how to remove flakes from cores by observing a human knapper; he also learned to use the sharp flakes to cut through a cord that held shut a reward box. After observing the procedure, Kanzi perfected his technique by trial and error. His initial solution, interestingly, was not to copy the demonstrator’s action, but to hurl the core onto a hard surface and then select the sharp pieces from the shattered remnants. He clearly understood the notion of breakage and its consequences. When experimenters padded the room, he then copied the hammering technique used by the knapper. From this experiment (and an earlier one by Wright(Wright 1972)), it is clear that fracturing stone is within the cognitive abilities of apes. However, Kanzi is not as adept as human knappers. " (A)s yet he does not seem to have mastered the concept of searching for acute angles on cores from which to detach flakes efficiently, or intentionally using flake scars on one flake of a core as striking platforms for removing flakes from another face"(Toth et al. 1993: 89). These abilities are basic to modern knapping and, more telling, are evident in the tools made two million years ago. Toth et al. suggest that this represents a significant cognitive development, though they do not specify just what cognitive ability may have evolved. Elsewhere (Wynn, Tierson et al. 1996)I have suggested that it may represent an evolutionary development in "spatial visualization," which is the ability to discriminate patterns in complex backgrounds. If true, this would represent a minor cognitive development, of interest primarily because it is a cognitive ability tied to tool manufacture and use. Kanzi is also not very accurate in delivering blows, and this is harder to assess. It could simply be a matter of biomechanical constraint (i.e., he does not have the necessary motor control), or it could result from an inability to organize action on the small spatial field of the core. It is the organization of such action, fossilized as patterns of flake scars, that developed significantly over the two million years following the first appearance of stone tools.

While apes can knap stone, they do not produce symmetries (at least not yet). The only possible example of symmetry produced by apes in the wild is the chimpanzee sleeping nest, which has a kind of radial symmetry that is produced when the individual reaches out from central position and pulls branches inward. Here the symmetry is a direct consequence of a motor habit pattern, and one need not posit some idea of symmetry (Wynn and McGrew 1989). There are no other ethological examples, at least to my knowledge. However, there has been a significant amount of research with captive apes, especially chimpanzees, including a fascinating literature about chimpanzee art and drawing, from which one can examine the ways apes arrange elements in space.

Work with ape art has been of two kinds. In the first, researchers present an ape with appropriate media (finger paints, brushes and paint, etc.) and encourage it to create. In the second, researchers control the productions by supplying paper with pre-dawn patterns. The former is the more "archaeological", in that researchers have not tried to coax particular pattern productions. Perhaps not surprisingly, these spontaneous productions are patterned primarily by motor patterns. Fan shapes are common, as are zig-zags produced by back and forth arm motion.

[FIG. 2]

Desmond Morris(Morris 1962), the most well-known researcher in ape art, thought that these productions may demonstrate a sense of balance, and tried to coax it out with a series of experiments using sheets with stimulus figures already printed on (Figure 1B), following the

earlier lead of Schiller(Schiller 1951). Morris’s work led to a number of subsequent experiments by others using similar techniques. The results have been enigmatic at best. Most chimpanzees presented with a figure that is offset from the center of the paper will mark on the opposite side, or on the figure itself (Fig. 2B). Morris suggested, cautiously, that this confirmed a notion of balance. Later Smith(Smith 1973) and Boysen(Boysen, Berntson et al. 1987) confirmed these results, but argued that the pattern resulted from the chimpanzee’s placing marks toward the center of the vacant space; balance was an accident.

It is hard to know what to make of this evidence. First, even with the few experimental subjects, there was a lot of individual variability. Indeed, each chimpanzee had an idiosyncratic approach to both the controlled and uncontrolled drawing. Second, most repetitive patterns resulted from repetitive motor actions. Nevertheless, the individuals did appear to place their marks non-randomly, and did attend to features of the visual field. Other, non-graphic, experiments have indicated that chimpanzees can be taught to select the central element of a linear array(Rohles and Devine 1967), so chimpanzees can clearly perceive patterns in which balance is a component. But they do not appear able to produce symmetrical patterns.

2.3 Tools of Early Hominids

2.3.1 Description

The earliest hominids left no archaeological record. Studies of blood chemistry and DNA indicate that humans and chimpanzees shared a common ancestor as recently as five million years ago. By four millions year ago the evolutionary split between hominids and the other African apes had occurred. There is fossil evidence for these early hominids, but it is fragmentary and more tantalizing than informative in regard to adaptive niches(Tattersall 2000). Between 4 million and 2.5 million years ago several different hominids lived in Africa. They differed from one another in habitat and adaptive niche, but shared the basic suite of hominid characteristics: bipedal locomotion, and relatively small canines and large molars. None had a particularly large brain (though slightly larger, relatively, than that of chimpanzees), and none left any archaeological traces. If any or all of these hominids made and used tools, as modern chimpanzees clearly do, then they have not been preserved. We can assume that tool use must have been in the repertoire of at least one of these hominids, only because it seems unlikely that stone tool manufacture could have developed without antecedents.

To date, the oldest reliably dated stone tools are 2.5 million years old (Harris 1983). These earliest stone tools exhibit no convincingly symmetrical patterns. Archaeologists assign these tools to a category termed "Oldowan," because of their first discovery at Olduvai Gorge in Tanzania. A better label was proposed several decades ago by Graham Clark(Clark 1977), who termed them a Mode 1 technology, a term based on technological characteristics, with no time-space implications. Mode 1 tools first appeared about 2.5 million years ago in what is today Ethiopia, and were the only kind of stone technology in evidence for the next one million years. After 1.5 million years ago, Mode 1 technologies continued to be produced in many areas and, indeed, were made into historic times. As such Mode 1 represents a common, "generic" stone tool industry. It was also the earliest

[FIG 3].

The emphasis of Mode 1 tools is on the edges(Toth 1985). Simple stone flakes can have very sharp edges, and are useful cutting tools without further modification. The cores from which the flakes were removed also have useful edges. These are not as sharp as the flakes, but the cores are usually heavier, and the result is a tool that can be used for chopping, crushing, and heavy cutting. Mode 1 tools exhibit little or no attention to the overall shape of the artifact. The only possible examples of a shaped tool occur in relatively late Oldowan assemblages, where there are a few flakes with trimmed projections (termed awls). Here a two-dimensional pattern of sorts has been imposed on the artifact, but it is a very "local" configuration, one that is tied to the nature of the edge itself.

The work of Stout et al. discussed earlier supports an emphasis on the spatial cognition required by basic kind stone knapping typical of these Mode 1 artifacts. Cognitive psychology supplies some more specific variables that are also applicable to the analysis of these early tools. Forty years ago Piaget and Inhelder (Piaget and Inhelder 1967) introduced basic topological notions in their analysis of children’s spatial ability, and these still have descriptive power. In particular, the relations of proximity, order and boundary are all required for the placing of trimming on Mode 1 tools. More recently, Linn and Petersen (Linn and Petersen 1986) have identified "spatial perception", the ability to detect features among complex backgrounds, as one of the four components of spatial cognition. This ability appears to be required when a knapper selects a platform with an appropriate angle for striking. What does not appear to be necessary for these tools is any kind of shape recognition or imagery. Basic flaking procedure and simple spatial relations are sufficient. The knappers imposed no overall shapes.

In this respect, at least, these early hominids were very ape-like. Indeed, when we expand our perspective to other features of tool making and use, we find that it was ape-like in most respects (Wynn and McGrew 1989). Yes, use of stone tools to butcher parts of animal carcasses obtained through scavenging was a novel component to the adaptation (Toth and Schick 1986; Potts 1988; Schick and Toth 1993), but at this point in hominid evolution it appears to have been merely a variant on the basic ape adaptive pattern, with no obvious leap in intellectual ability required. Indeed, there is no compelling archaeological reason to grant tool making any special place in the selective forces directing the first three million years of human cognitive evolution. But sometime after two million years ago the situation changed.

About 1.4 million years ago hominids in East Africa, presumably Homo erectus, began making large stone tools with an overall two-dimensional shape. Many (but not all) of these "bifaces" were made by first detaching a very large flake from a boulder-sized core using a two-handed hammering technique (Jones 1981). The knapper then modified this large flake by trimming around the margins (usually onto both faces of the flake, hence the term biface). The uses to which these tools were put are unknown, though experimental evidence indicates that they can be effective butchery tools (Toth and Schick 1986). Archaeologists recognize two types of biface, the handaxe and the cleaver. Handaxes have a point or tip, and cleavers have a transverse "bit" that consists of an untrimmed portion of edge oriented perpendicular to the long axis of the tool.

Both varieties of biface can have reflectional symmetry, and it is primarily this symmetry that produces the overall shape. However, not all bifaces of this age are nicely symmetrical, and even the nicest examples look crude compared to the symmetry of later tools. Are we justified in attributing some kind of symmetry concept to the knapper?

[FIG 4]

Might not the symmetry lie only with archaeologist, who "reads" what was in no way intended by the knapper? This is a knotty problem that has become the center of an interesting, if parochial, controversy among cognitive archaeologists (Noble and Davidson 1996). Most archaeologists, myself included, argue that the symmetry is real. First, the most symmetrical examples are also the most extensively trimmed, indicating that the knapper devoted more time to production. Second, and more telling, on some bifaces the trimming mirrors a natural shape on the opposite edge. Such artifacts do not have the best symmetry, but the economy of means by which the symmetry was achieved reveals that some concept of mirroring must have guided the knapper.

[FIG 5]

In addition to handaxes and cleavers, a third variety of biface occurs in low numbers in some sites in this time period. These are "discoids", so called because of their round shapes. Like the other bifacial tools, the nicest, in this case the roundest, are also the most extensively modified. Here again we can recognize symmetry, in this case radial rather than reflection.

[FIG 6]

In most respects the cognitive requirements of these early bifaces resemble those of the earlier (and presumably antecedent) Mode 1 artifacts. But the symmetry presents a puzzle for cognitive interpretation. There are at least three possibilities:

1) The symmetry (and regular radii) are purely a consequence of a technique of manufacture using large flakes as blanks. The placement of trimming on some pieces argues against this, but the absence of congruency means that the symmetry is always crude and, for many archaeologists, unconvincing (Noble and Davidson 1996). Any cognitive significance would have to lie in the techniques of blank production. Although the two handed hammering technique (Jones 1981) clearly qualifies as an invention, its cognitive prerequisites seem no different from those of other direct percussion techniques used in Mode 1 technologies.

2) The symmetry was intended, but was not "new." Rather it is a pattern that is salient in the shape recognition repertoire of apes in general. What was new was the imposition of this shape on modified objects, something other apes never do.

3) Symmetry was a new acquisition in the shape recognition repertoire of these hominids and was applied to stone tools.

Conservatism inclines me towards the second hypothesis. But even if symmetry in pattern recognition is old, there was still a cognitively significant development associated with these bifaces. The stone knappers produced a symmetry by mirroring or reflecting the shapes from one lateral edge to the other. True, the edges are not exact mirrors. They are rarely if ever congruent in a modern geometric sense, but they are inversions of a two-dimensional shape. It is not even necessary that a particular overall shape have existed as an image prior to manufacture. The knapper could simply have mirrored one of the edges naturally provided him or her. In such a case the knapper would need to invert a shape. More significant, the knapper had to ignore part of the shape of the original large flake in order to impose a symmetrical edge. This is a kind of frame independence, the ability to see past the constraints imposed by a spatial array (Linn and Petersen 1986). The discoids suggest that the knappers were also able to employ a notion of spatial amount, in this case a diameter. The knappers trimmed the tool until all of the diameters were roughly equal. While not an abstract quantity like an inch, a diameter is nevertheless a spatial amount, albeit local and limited. But what is most significant is that these biface knappers incorporated a shape component into the knapping problem. This shape component need not have been an abstract concept. It could simply have been shape "recognition," matching to unimodal representation (Kosslyn 1999), in this case reflectional symmetry. Such a unimodal representation is almost certainly in the shape recognition repertoire of apes in general. What is significant here is its manifestation in the otherwise spatial task of knapping.

This new development required coordination of spatial abilities with a previously separate cognitive component (or neural network in Kosslyn’s sense(Kosslyn 1994)), that of shape recognition.

The imposition of shape is a feature of virtually all human material culture. But the first time it ever occurred was with these early bifaces. Prior to the appearance of bifaces, stone knappers attended to the configuration of edges and to size. The earlier Mode 1 tools were arguably an ad hoc technology (Wynn 1981; Isaac 1984; Toth 1985) made for immediate use. It is unlikely that they existed as "tools" in the minds of the knappers. But tools, in the guise of bifaces, almost certainly did exist as a category in the mind of Homo erectus (Wynn 1993; Wynn, Tierson et al. 1996).

Three developments in hominid imposed symmetry appear in the archaeological record sometime after 500,000 years ago. These are: 1) congruency; 2) three-dimensional symmetries; and 3) broken symmetry.

While the reflectional symmetry of early bifaces was rough and imprecise, the symmetry of late examples clearly suggests attention to congruency. The mirrored sides are not just qualitative reversals, but quantitative duplicates, at least to the degree that this is possible given the constraints of stone knapping. Many, but certainly not all, late handaxes and cleavers present such congruent symmetries, and this is one of the features that makes them so attractive to us. Such a symmetry was not limited to a single shape. Late bifaces demonstrate a considerable amount of variability in overall plan shape. Some are long and narrow, others short and broad. Some have distinct shoulders, while others are almost round. Although there is some evidence that this variability was regional(Wynn and Tierson 1990), much of it is related to raw material, and much appears to have been idiosyncratic. But in almost every assemblage of this time period there will be a few bifaces with fine congruent symmetry, whatever the overall shape.

The second development in symmetry was the appearance of reflectional symmetry in three dimensions. Many of these bifaces have reflectional symmetry in profile as well as in plan. In the finest examples this symmetry extends to all of the cross-sections of the artifacts, including cross-sections oblique to the major axes, as we would define them.

[FIG 7]

Once again, this feature is not universally true, and many, many bifaces do not have it, but it is present on at least a few artifacts from most assemblages.

The third development in symmetry was the appearance of broken symmetry. Here a symmetrical pattern appears to have been intentionally altered into a non-symmetrical but nevertheless regular shape. Several cleavers from the Tanzanian site of Isimila appear "bent," as if the whole plan symmetry, including the midline, had been warped into a series of curved, parallel lines. These are invariably extensively modified artifacts, whose cross-sections are symmetrical, and the pattern is almost certainly the result of intention.

[Figure 8]

A better known example is the twisted profile, or "S-twist", handaxe. The artifacts give the appearance of having been twisted around the central pole. The result is an S-shape to the lateral edges, as seen in profile.

[Figure 9].

Again, these are extensively modified artifacts and we must conclude, I think, that the pattern is the result of intention.

It is not possible to date these developments in symmetry precisely. Archaeological systematics place all of the examples in the late Acheulean (sometimes on morphological grounds alone, which leads to a circular argument). All were probably made after 500,000 years ago, perhaps even after 400,000 years ago. The Isimila artifacts, for example, date to between 170,000 and 330,000 years ago(Howell, Kleindienst et al. 1972). The twisted profile handaxes are probably no earlier than 350,000, and most may be much later. Although 300,000 years is a long time in a historical framework, it represents only the final 12% of technological evolution.

Several caveats complicate interpretation of these three developments. One is the problem of individual skill; some prehistoric stone knappers must have been more adept than others and better able to achieve congruent, three-dimensional symmetries in the intractable medium of stone. We have no way of knowing how common highly skilled knappers were. A second caveat is raw material. Some stone is much easier to work that others. I do not think it is entirely coincidence that twisted profile handaxes are invariably made of flint or obsidian, two of the most prized knapping materials. On the other hand, raw material is not as tyrannical as one might think. The "bent" cleavers from Isimila are made of granite.

The imposition of three dimensional, congruent symmetry probably depended on cognitive abilities not possessed by the first biface makers. The cognitive psychological literature suggests some possibilities. The first requirement would appear to be the ability to coordinate perspectives. While flaking the artifact, the knapper has only one point of view. This is adequate to control edge shape, and perhaps even two-dimensional symmetry, but to produce an artifact with three-dimensional symmetry one must somehow "hold in mind" viewpoints that are not available at that moment, and for the finest symmetries viewpoints that are not directly available at all (oblique cross-sections). The knappers must have understood the consequences of their actions for the shape of the artifact as it appeared from these other perspectives. Such manipulations are akin to "allocentric perception" recognized by psychologists (Silverman, Choi et al. 2000), and used in image manipulation tasks such as mental rotation. It is likely that these hominids were able to manipulate mental images of objects. Again, archaeological bias forces a conservative analysis; however, no one has proposed a convincing simpler alternative to this one. Application of a simple flaking procedure, without any image manipulation, could not have produced the kinds of three-dimensional symmetries evident on these artifacts. The second requirement, congruency, is clearly spatial in the narrow sense of perceiving and imaging spatial quantity. As we have seen, basic knapping is largely a spatial problem, and was from the beginning. What is new here is the application of metric spatial relations to a problem of shape. Simple unimodal shape "recognition" would not have been enough. The sophistication of this symmetrical pattern suggests that shape "identification" is required. "When we recognize something, we know only that we have perceived it before, that it is familiar [re. early handaxes above]; when we identify something, we know it has a certain name, belongs to specific categories, is found in certain locales, and so forth"(Kosslyn 1999):1284). These handaxes were almost certainly categories, and categories are abstract, multi-modal, and rely on associative memory. As such they reside in declarative memory, which "...requires associative links between several types of information that are stored in different areas"(Ungerleider 1995):773).

These hominids could manipulate perspectives and spatial quantity, produce congruent symmetries, and even distort these principles to achieve striking visual effects. It is fair, I think, to attribute an intuitive Euclidean concept of space to these stone knappers. A Euclidean sense of space is one of three-dimensional positions. While the human life-world is certainly organized this way, and we and other primates clearly perceive dimensional space, it is quite another thing to employ cognitive mechanisms that understand space in this way, and which can be used to organize action. Such a mechanism, or mechanisms, underpin our most sophisticated everyday navigational and mapping skills.

The examples I have used thus far have all been knapped stone artifacts. While symmetry clearly can be and was imposed on many knapped stone artifacts, the medium is not ideal for the imposition of form. It is not plastic, and shaping can only be done by subtraction. Indeed, after the appearance of the symmetrical patterns just discussed, no subsequent developments in symmetry can be recognized in knapped stone. There were developments in technique, and perhaps skill, but the symmetries imposed on even very recent stone tools are no more elaborate than those imposed on 300,000 year old handaxes. As a consequence we must turn to other materials.

Artifacts made of other materials -- bone, antler, skin, wood, fiber, etc. -- were undoubtedly part of the technical repertoire of many early hominids (though see(Mithen 1996) for a counter suggestion). Because such materials are far more perishable than stone, the archaeological record contains few of them until relatively late in prehistory. There are a few examples that almost certainly pre-date 100,000 years ago, but all are controversial, either as to age, or as to significance. One is an pebble from the Hungarian site of Tata, on which someone engraved a line perpendicular to a natural crack(Bednarik 1995). While one might be tempted to argue from it that the maker had some notion of rotation, or radial symmetry, this is too heavy an interpretive weight to be born by a single, isolated artifact. More to the point, even if true, this would tell us little more about symmetry than is supplied by contemporary bifaces. However, it would be symmetry in a new context, a fact which if confirmed would have possible implications for cognitive evolution.

It is not until very close to the present, indeed after 100,000 years ago that the archaeological record presents extensive evidence of artifacts made of perishable materials. Some archaeologists see this timing as entirely a reflection of preservation; others see it as evidence of new behaviors and abilities. The earliest such evidence is African and dates from between 50 and 90,000 years ago(Yellen, Brooks et al. 1995; Klein 2000)). These are worked bone points from a site in the eastern Congo. While these artifacts are quite important to several current arguments about prehistory, they reveal nothing new in regard to hominid imposed symmetry. The European Upper Palaeolithic provides the best documented examples of hominid imposed symmetries for the time period between 40,000-10,000 years ago. Here we find extensive evidence of symmetry in materials other than stone.

[Fig. 10]

Perhaps most widely known are cave paintings of Franco-Cantabrian Art, especially in compositions that are about 15,000 years old. Here we can see possible symmetries as patterns of elements in a composition, not just inherent in a single object. They appear to have resulted from the application of a compositional rule. As such, they do not inform us specifically about spatial or shape cognition and are outside the scope of this discussion.

I suggested at the beginning of this article that archaeology can make two important contributions to the study of the evolution of human cognition: the timing of certain developments and a description of the evolutionary context in which these developments occurred. The sequence of development of hominid imposed symmetries just summarized allows us to do both of these things.

The development of artifactual symmetry was not slow and continuous. Instead, the archaeological evidence suggests that there were two episodes of development, separated by as much as one million years. During the first, hominids developed the ability to impose shape on artifacts, an ability undocumented for any living apes. In doing this early Homo erectus employed cognitive abilities in frame independence, mirroring, making simple judgments of spatial quantity, and coordination of shape recognition (symmetry) with the spatial requirements of basic stone knapping. There may have been others, but these are the ones evident in the archaeological record. This development occurred early in the evolution of the genus Homo, certainly by 1.4 million years ago. The second episode of development occurred much later and consisted of the acquisition of a modern Euclidean set of spatial understandings. Specific abilities evident from the symmetrical handaxes include congruency, three-dimensional shapes, and coordination of perspectives. The date of this development appears to correlate with the evolutionary transition from Homo erectus to Archaic Homo sapiens.

This timing of developments has implications for human cognitive evolution. First, the initial hominid adaptation (4.5-1.5 mya) apparently included a basic ape-like understanding of space and shape. A generalized ape repertoire of spatial concepts was adequate for this earliest of hominid adaptive niches, including the first manufacture and use of stone tools. A distinctive set of spatial/shape abilities did not appear until relatively late in hominid evolution, after the appearance of Homo erectus. Second, because these two later episodes of cognitive development were discontinuous, and indeed rather far from another in time, it is unlikely that they occurred in response to the same selective factors. Whatever selected for the spatial/shape abilities of early Homo erectus probably did not select for the Euclidean abilities that emerged one million years later. But perhaps the most important implication that the development of artifactual symmetry has for the understanding of human shape and space cognition in general, and not just its developmental sequence, is that even the more recent developments occurred in the very remote past. In terms of shape and spatial thinking we have not just Stone Age minds, we have Lower Palaeolithic minds.

3.2 Evolutionary Context

Evolutionary context is the second body of information archaeology can provide the study of the evolution of cognition. While it is well and good to describe a sequence of development, it would also be good to answer the questions of how, and perhaps why. In evolutionary science this amounts to answering the question of selection. What selected for these abilities? Evolutionary psychologists (Barkow, Cosmides et al. 1992; Bock and Cardew 1997) answer this question by looking at evidence for adaptive design, on the assumption that past selection is preserved in the modern architecture of the cognitive mechanisms. Paleoanthropologists, and archaeologists in particular, are suspicious of such reliance, and prefer to invoke the actual context of development to help identify possible selective agents. While our knowledge of the conditions of the evolutionary past is fragmentary and lacking in detail, it is still an account of actual prevailing conditions, not a reconstruction based on presumed selective pressures.

Hominid fossils and the archaeological record constitute the primary evidence for the context of cognitive evolution, supported by a large body of methods used for dating and for reconstructing the physical environment. Hominid fossils provide some direct evidence of cognitive evolution in the guise of brain size and shape. At least at our present level of understanding this does not lead to persuasive arguments about specific abilities, but it can identify times of brain evolution in general, which can support arguments derived from other evidence. Hominid fossils can also inform us about other evolutionary developments in anatomy, which human paleontologists have used successfully to document changes in diet, nutrition, locomotion, heat and cold adaptation, levels of physical stress, and other aspects of adaptive niche that are directly relevant to the context of cognitive evolution. Archaeological evidence, because it is far more abundant than fossils, informs about geographic distribution, habitat use, specific dietary components, geographic range (via raw material transport), and cultural solutions to problems (fire, weapons, boats, etc.), in addition to the evidence for hominid cognitive abilities. Together the fossil and archaeological evidence provide a reliable, if incomplete, picture of the past, including the two time periods in which the major developments in artifactual symmetry occurred.

We know much less about the first episode of development than the second. 1.4 million years ago, the time of the first biface industries with their evidence for the imposition of symmetry and concomitant modest developments in spatial thinking, was also the time of Homo erectus. Indeed, the first Homo erectus (a.k.a. Homo ergaster (Tattersall 2000)) had appeared in Africa (and perhaps elsewhere) several hundred thousand years earlier, so we cannot make a simple equation between Homo erectus and biface technology. Luckily, one of the most spectacular fossil finds for all of human evolution is an African Homo erectus from this time period. The Nariokotome Homo erectus is an almost complete skeleton of a youth who died about 1.55 million years ago (Walker and Leakey 1993). The completeness of the skeleton allows a more detailed discussion of life history and physiological adaptation than is possible with fragmentary remains. The youth was male, about 11 years old, stood about 160cm (63") at time of death (estimates of adult stature for this individual are 185cm [73"]), had a tall, thin build, and evidence of strenuous physical activity. His brain size was about 880cc, and the endocasts demonstrate the same left parietal and right frontal petalia typical of humans but not of apes. His thoracic spinal diameter was smaller than that of modern humans of similar size, and he had a very small pelvic opening, even for a thin male. This anatomy suggests several important things about his physiology. He had the ideal body type for heat dissipation. Added to the evidence for strenuous activity, this suggests that exertion in hot conditions was common. Earlier hominids had been largely woodland creatures who focused most of their activity close to standing water. Nariokotome had the anatomy to exploit an open tropical grassland adaptive niche. While the brain size of Nariokotome was larger than earlier hominids, so was his body size; there was only a small increase in relative brain size (compared to, say, Homo habilis). Despite the modern overall shape of the brain, the thoracic spinal diameter (and by extension the number of nerve bodies enervating the diaphragm muscles) suggests that rapid articulate speech was not in Nariokotome’s repertoire.

In sum, Nariokotome suggests that the Homo erectus niche was significantly different from that of earlier hominids, including earlier Homo. It is not clear from the cranial capacity that a significant increase in braininess accompanied this adaptive shift. There is no good reason to think Homo erectus had speech, at least in a modern sense (Wynn 1998). However, the niche shift itself was very significant, and is corroborated by the archaeological evidence.

Archaeological sites from this time period are less informative about hominid activity than many earlier sites. This seeming paradox results from the typical sedimentary context of the sites. Most early biface sites have been found in stream deposits, rather than the lake shore deposits typical of earlier sites. These "high energy" environments move objects differentially, including bones and tools. In effect they destroy the natural associations on which archaeologists rely. Running water also modifies bone, and to a lesser extent stone tools. The unfortunate result is that archaeologists have few direct remains of activity other than the stone tools themselves. However, there are enough sites dating to this time period to allow archaeologists to assess geographic distribution and environmental context, both of which suggest that a significant change in niche had occurred.

Homo erectus left stone tools in stream beds because he had moved away from permanent standing water. Archaeologists presume that the channels of ephemeral streams, or the banks of permanent streams, became one of the preferred activity locales. Given the absence of associated materials, we cannot determine just what these activities were; only the selection of locale is apparent. However, this fits nicely with the "body cooling" anatomy of Nariokotome. On open savannas, stream channels often support the only stands of trees (in addition to water). Archaeologists have also found African biface sites at higher altitudes than earlier tools sites (Cachel and Harris 1995), and, finally, there are early biface sites outside of tropical Africa. The best known is Ubeidiya in Israel, which in most respects resembles early biface sites in Africa (Bar Yosef 1987).

This archaeological evidence presents a picture of Homo erectus as an expansionistic species who invaded new and varied environments. Cachel and Harris (Cachel and Harris 1995) suggest it was a "weed species" - never numerous individually but able to invade new habitats very rapidly. Given the clear reliance on tools, Homo erectus’ niche was at least partly technological. Control of fire may also have been a component (James 1989). Evidence from this time period at the south African site of Swartkrans includes convincing evidence of the control of fire(Brain and Sillen 1988). While control of fire may have little cognitive significance (McGrew 1989), the importance to adaptive niche may have been profound in terms both of warmth and predator protection. We have little direct evidence for diet. >From experimental studies we know that bifaces could be effective butchery tools, but there are no obvious projectiles and no evidence for greater reliance on meat. There is, in fact, no compelling evidence for hunting.

It is not clear from this picture just what might have selected for the development in cognitive abilities evident in artifactual symmetry. At the outset we can consider the possibility that natural selection acted directly on the hominid ability to recognize and conceive of symmetry, which is, after all the pattern that is so salient in the archaeological record. What might the perceptual saliency of symmetry have been good for? There is considerable evidence that body symmetry is, in fact, related to reproductive success for males(Gangestad 1997). According to Gangestad, observable phenotypic asymmetry (away from the reflectional symmetry coded genetically) correlates with developmental stress, so that asymmetry marks lower health. If a potential mate could detect this, he or she could avoid a reproductively costly (in an evolutionary sense) mating. But presumably this is true generally for vertebrates, and not just for humans. Perhaps symmetry gained added importance as a clue to general health when hominids lost thick body hair. Condition of coat is also a good indicator of general health, and its absence may have led to selection for a heightened ability to detect variations away from symmetry. But the real question here is why would Homo erectus have imposed symmetry on artifacts? Could artifacts have come to play a role in mate selection? Could symmetry have become so salient a pattern for mate assessment that it intruded into other shape recognition domains? Here the saliency of symmetry has been transferred out of the domain of the phenotypic to that of cultural signaling, but the selective advantage is the same. In this scenario both the ability to detect and produce symmetry would have had reproductive consequences. Unfortunately, it is difficult to see how such a hypothesis could be tested. It is provocative only because of the known role of symmetry in mate selection.

Given the change in niche associated with early Homo erectus, with the accompanying increase in range and the pioneer aspects of the adaptation, it is tempting to posit selection for spatial cognition via navigational ability. Judging spatial quantity would be a useful skill, for example. However, while matching diameters on tools and judging distances to water are both matters of spatial quantity, it is not clear that they use identical cognitive mechanisms. Indeed, neurological research suggests that relationships in "near" and "far" space are handled somewhat differently by the brain (Marshall and Fink 2001), though there does appear to be some correlation (Silverman, Choi et al. 2000). The temporal association of territory expansion with developments is shape and spatial cognition is provocative, but hardly conclusive. Given the emphasis in some literature on sexual division of labor (Eals and Silverman 1994; Silverman, Choi et al. 2000), it is also important to reiterate that paleoanthropologists know nothing about division of labor in this time period.

Even though the contextual evidence provides no leading candidate for selective agent, it does describe an adaptive milieu of relevance. Early Homo erectus was not much like a modern hunter-gatherer. There is no evidence for human-like foraging systems or social groups (the probable absence of speech would itself obviate the latter). There is not even any convincing evidence of hunting with projectiles, a favorite of several authors (Calvin 1993). Nothing in the contextual evidence warrants direct analogy to the adaptive problems of modern human foragers. The challenge that Homo erectus presents to paleoanthropologists, and other students of human evolution, is that there are no living analogs. There is no more reason to invoke a human model than a chimpanzee model, or neither.

Paleoanthropologists’ knowledge of evolutionary context is much better for the time period associated with the appearance of three-dimensional symmetries, congruency, and multiple perspectives. These abilities were clearly in place by 300,000 years ago, and probably by 500,000. Our knowledge is better partly because this time period was much closer to the present (though still remote), but also because Homo had expanded into Europe. It is true that some temperate environments have good preservation, but the primary archaeological effect of this expansion is that Homo moved into an area where a great deal of modern archaeology has been done. Africa may be the home of mankind, but Europe is the home of archaeology. Based largely on European sites it is possible to draw an outline sketch of the behavioral/cultural context of daily life.

The peopling of Europe is itself a fascinating topic. Some argue that Europe was occupied relatively late, after 500,000 year ago (Roebroeks, Conard et al. 1992). Earlier sites are certainly scarce and often controversial. However, there are sites in Italy (Isernia(Cremaschi and Peretto 1988), Ceprano(Ascenzi, Mallegni et al. 2000)) and Spain (Atapuerca(Bermudez de Castro, Arsuaga et al. 1997)) that provide strong evidence for the presence of Homo erectus (now attributed by some to Homo antecessor) prior to 500,000. In many respects these resemble earlier Homo erectus sites - poor geological context and little to go on. There is little doubt that after 500,000 the record is better and includes the peopling of northern Europe. There are informative sites in England, Germany, France, and the Netherlands. Some archaeologists have suggested that this represents an adaptive breakthrough, an evolutionary development that opened up harsher environments (Roebroeks, Conard et al. 1992). These first colonists in northern Europe did make bifaces, and their bifaces required all of the cognitive abilities identified earlier in this paper. The association is provocative, and suggests that some evolutionary development in cognition may have been partially responsible for this breakthrough. Of course, similar artifacts appeared in Africa, so the breakthrough cannot have been specific to Europe. But what specifically might have selected for cognitive abilities evident in the fine three-dimensional symmetries of bifaces? The archaeological record does provide some clues.

The move into northern Europe (and China) may seem a minor extension of the much more dramatic expansion of Homo erectus 1,000,000 years earlier. However, it may well have required some significant changes in adaptation. Northern European climate during the Pleistocene was in more or less constant flux, with cold glacial periods alternating with warmer, forested, interglacial periods. During periods of maximum cold northern Europe was uninhabitable, and indeed even the anatomically modern of humans of 18,000 years ago were forced to the south. But after 500,000 we have evidence of biface sites in northern Europe during some of the warmer episodes embedded in glacial periods. The environment during these periods was more open and less heavily wooded than today, but also cooler than today. The adaptive problems posed by such an environment are fairly well known. Compared to warmer environments, even in southern Europe, there would have been fewer edible plant species, and a concomitant requirement for increased reliance on animals. Then there obvious problems of keeping warm, including the likely necessity of controlling and probably even making fire. In effect, these northern temperate environments "pushed the envelope" of Homo’s adaptation. But here our European bias risks misleading. We see it clearly because we have the sites. We know, however, that comparable technological developments occurred in Africa, the Near East, and China, and it is unlikely that Europe was in any way central (indeed cultural backwater is more likely), so what we may be seeing is abilities that evolved elsewhere applied to a European problem.

The fossil remains from this time period present a confusing picture. In some areas of the world, Asia in particular, Homo erectus was clearly still present. In Africa the prevailing fossil type appears to have been a larger brained form that still retained many Homo erectus like features of the face. Europe is the biggest puzzle. There are few clear Homo erectus fossils (the Mauer mandible is a probable example (Rightmire 1992), as is the Ceprano calvaria(Ascenzi, Mallegni et al. 2000)) but most fossils attributed to the taxon have some at least some sapient-like features. Rightmire (Rightmire 1998) now favors a return to the use of Homo heidlebergensis for these forms. The key issue for the present argument is that this was a time of evolutionary change in anatomy, as well as technology and cognition. We cannot yet describe the complex evolutionary patterns, but we do know the long period of evolutionary stasis that was Homo erectus was coming to an end.

By this time it is almost certain that these Homo hunted large mammals. Many sites have included the association of stone tools with animal bone (Hoxne(Singer, Gladfelter et al. 1993), Boxgrove (Wenban-Smith 1989; Roberts, Stringer et al. 1994; Roberts and Parfitt 1999)), and in some cases like Hoxne the range of body parts present suggests that at least some of these animals had been hunted, not just scavenged (Binford 1985). Until recently the only evidence of hunting technique was an enigmatic sharpened stick from Clacton-on-Sea that could have been a spear tip (Oakley, Andrews et al. 1977). In 1996, the German site of Schoeningen dramatically confirmed this interpretation. Here Thieme (Thieme 1997) and crew excavated three complete spears carved out of spruce, each over two meters in length. The spears were in direct association with the bones of horses. The center of gravity of each of these spears was situated slightly toward the tip, much as it is in modern javelins, and Thieme argues that Homo must have designed these spears for throwing. If true, this suggests a relationship between design and use that has technological and perhaps even cognitive implications. But what bears more on the issue at hand is that hunting with spears was obviously a component of the behavioral/cultural context of these Homo, a component that calls up modern human analogs.

The archaeological record does not provide much direct evidence for group organization, or even size, at least not for this time period. For more recent times the size and debris patterns of structures provides much useful evidence of social organization, but such patterns degrade rapidly and in only the most ideal conditions survive for tens of thousands of years, let alone half a million. All of the possible campsites from the time period of interest here are controversial. Thirty years ago de Lumley (Lumley and Boone 1976) presented a dramatic argument for huts, seasonal occupation, and reuse at the French site of Terra Amata, an interpretation that still survives in textbooks(Turnbaugh, Jurmain et al. 1999). The site has never been published in detail, and the one independent analysis cast considerable doubt on de Lumley’s interpretation(Villa 1983). At best, there is evidence for a few post holes, stone blocks, stone knapping debris, and shallow hearths scooped from the sand. These may be the remains of a flimsy shelter, and would certainly fit into a reconstruction of a small hunter gatherer band. Unfortunately, this optimistic picture is at least premature, and probably unwarranted. Villa’s analysis indicates that the integrity of the site is much lower than presented by de Lumley. It is in reality an accumulation of cultural debris that has been moved and altered by natural processes. Yes, this provides evidence of hominid activity, but not of a coherent campsite with multiple activities and an artificial structure. There are no better examples until much later. Indeed, Gamble (Gamble 1994; Gamble 1999) argues that this absence of campsites is an accurate reflection of the life ways of early Homo sapiens, and that coherent long-term or multiple use campsites were not part of the adaptive niche. If true, these early Homo sapiens were not like modern hunters and gatherers.

There are other ways in which they were not modern. As familiar as some of this evidence is, there is a striking piece of modern behavior that was entirely missing. We have no convincing evidence of art, or personal ornamentation, or anything that clearly was an artifact of symbolic culture. Many sites do have lumps of red ochre, some of which had been scraped or ground. A few sites have enigmatic scratched bones. None of this constitutes indisputable evidence that these hunters and gatherers used any material symbols, of any kind whatsoever. Compared to the abundant use of such items in sites post-dating 50,000 this absence is telling.

In sum, the archaeological evidence indicates that by 400,000 years ago Homo was a hunter gatherer who had invaded new, more hostile environments, but who did not invest in symbolic artifacts. Despite similarities to modern hunters and gatherers, these early Homo sapiens were different in many respects.

What might have selected for the cognitive abilities required for three-dimensional, congruent symmetries? Again, mate selection is a possibility, this time by way of technological skill. An individual who could produce a more regular (symmetrical) artifact would be cuing his or her skill, and worth as a potential mate. Other things being equal, the stone knapper who produced the fine three-dimensional bifaces was smarter and more capable, with better genes, than one who couldn’t. Especially if knapping skill correlated with other technological abilities this would be one means of identifying mates with future potential as providers. Kohn and Mithen(Kohn and Mithen 1999) take this argument even farther, framing it terms of sexual selection and emphasizing abilities other than technological.

Modern people certainly do use material culture to mark their individual success, and it is perhaps not far fetched to extend this behavior into the past, perhaps even to the time of late Homo erectus or early Homo sapiens.

As second possibility is that selection operated on enhanced spatial and or shape cognition, with artifactual symmetry being just one consequence. Given the co-occurrence of hunting and gathering and modern spatial thinking in the paleoanthropological record, this hypothesis suggests that they are somehow linked. What selective advantage could congruency, three dimensional symmetries, and image manipulation bestow on a hunter-gatherer? William Calvin (Calvin 1993) has long argued that aimed throwing was a key to cognitive evolution. While I find his argument that bifaces were projectiles far from convincing (see also (Whittaker and McCall 2001)), the Schoeningen spears may well have been, indeed probably were, projectiles. That there is a spatial component to accurate throwing seems beyond question. Calvin himself emphasizes the importance of timed release and the computational problems of hitting moving targets. Would any of this select for image manipulation, congruency, and so on? It is hard to see how, unless ability to estimate distance to target selected for abilities in judging all spatial quantities (e.g., congruent symmetries). The selective agent, throwing, just does not match up well with the documented abilities.

Navigation is again an alternative, and one favored by many psychologists (e.g. (Gaulin and Hoffman 1988; Moffat, Hampson et al. 1998); (Dabbs, Chang et al. 1998; Silverman, Choi et al. 2000)). While route following using sequential landmarks can work, using basic topological notions like those known for chimpanzees and early hominids, it is difficult to conceive of and follow novel routes without a Euclidean conception of location. Arguably hunting, especially varieties invoking long distance travel, herd following, or intercept techniques would favor Euclidean conceptions of space. There is now experimental evidence documenting a correlation between navigational skill and the standard psychometric measures of spatial cognition like mental rotation (Moffat, Hampson et al. 1998; Silverman, Choi et al. 2000), though recall that "near" and "far" space are not handled identically by the brain(Marshall and Fink 2001). This specific selective hypothesis, then, is a better fit than the throwing hypothesis. Of course, navigation skill might have been unrelated to hunting per se, and instead tied to mate searching (Gaulin and Hoffman 1988) or any long distance travel. What is provocative is the correlation between the earliest evidence for large scale hunting and Euclidean spatial relations, as represented by fine three dimensional symmetries. While the correlation between this development in spatial thinking and navigation is provocative, it does have two weaknesses. First, many animals are fine navigators without relying on the enhanced spatial abilities in question. Of course, what we are seeing here is the hominid solution to navigation problems, so I am not too troubled by this objection. The second objection is more bothersome. When modern people navigate, they rarely use the spatial abilities in question. For example, when modern hunters and gatherers move across the landscape they use established paths and routes, often following water ways or animal trails (Baluchet 1992; Gamble 1999). The geometric underpinning of such navigation is largely topological, and does not rely on the kind of spatial abilities evident in the stone tools. Yes, it is possible to imagine a form of navigation that relies on such abilities, but this does not seem to be the way people actually move about. Unless there is a compelling reason to think that modern hunter-gatherers rely on Euclidean spatial relations to navigate, it will remain a weak hypothesis.

Of course, these spatial and shape abilities may not have been directly selected for at all. They may be by-products of natural selection operating on other cognitive mechanisms. For example, if Kosslyn’s (Kosslyn 1994) (Kosslyn, Thompson et al. 1998)characterization of mental imaging is accurate, the key development may have been in central processing rather than the more encapsulated shape recognition system or spatial assessment system. These are relatively discrete neural networks that reside in different parts of the brain (one temporal lobe, one parietal). In order for someone to conceive of congruency, and perhaps alternative perspectives, the two outputs must be coordinated, and this coordination appears to happen in the association areas of the frontal lobe. Here the evolutionary development would be in the area of association and central processing, and there is no reason for selection to have been for shape recognition or spatial ability per se. In other words, the archaeological evidence for the development of three dimensional, congruent symmetries, may inform us about developments in more general cognitive abilities, not just a narrowly encapsulated module of spatial thinking.

4. Conclusion

The archaeological record of symmetry reveals two of the times at which significant developments in hominid cognition occurred. The first, a million and a half years ago, encompassed cognitive developments necessary to the imposition of shape on artifacts, the coordination of shape recognition (symmetry) and spatial thinking (stone knapping) being the most salient. This evolutionary development was associated with Homo erectus, and the appearance of the first hominid adaptation that was clearly outside the range of an ape adaptive grade. These Homo erectus were not, however, like modern hunter/gatherers in any significant sense; indeed, there are no appropriate analogs living today, and the precise agents selecting for these cognitive abilities are not apparent.

The second episode evident from artifactual symmetries occurred a million years later and encompassed the development of modern Euclidean understandings and manipulations of shape and space. This was the also time of the transition from Homo erectus to Archaic Homo sapiens. The appearance of large mammal hunting in the contemporary archaeological record lends some support to evolutionary psychological arguments that hunting may have selected for features of human spatial cognition, either by way of projectile use or navigation. However, given the range of evidence documenting the appearance of many features of hunting and gathering at this time -- not just spatial thinking -- it is perhaps simpler to posit a few developments in associative abilities than a raft of specific cognitive mechanisms. It is also important to reiterate that despite being Homo sapiens, these were not modern hunters and gatherers. They lacked the rich symbolic milieu on which modern humans, including hunters and gatherers, rely.

Archaeology cannot itself resolve many of the controversies raised by the evidence. Questions concerning the cognitive and neural bases of the actions preserved in the archaeological record must be answered in studies of modern cognition. Archaeology can point to the times and contexts of cognitive evolution, but cannot itself illuminate the workings of the human mind. A comprehensive approach to cognitive evolution must therefore be multidisciplinary.

Ascenzi, A., F. Mallegni, et al. (2000). "A re-appraisal of Ceprano calvaria affinities with Homo erectus, after the new reconstruction." Journal of Human Evolution 39: 443-450.

Baluchet, S. (1992). Spatial mobility and access to resources among the African Pygmies. Mobility and Territoriality: Social and Spatial Boundaries Among Foragers, Fishers, Pastoralists, and Peripatetics. M. Casimir and A. Rao. Oxford, Berg.

Bar Yosef, O. (1987). "Pleistocene connexions between Africa and Southwest Asia: an archaeological perspective." African Archaeological Review 5: 29-38.

Barkow, J., L. Cosmides, et al., Eds. (1992). The Adapted Mind: Evolutionary Psychology and the Generation of Culture. New York, Oxford University Press.

Bateson, G. (1972). A re-examination of "Bateson's Rule". Steps to an Ecology of Mind. G. Bateson. New York, Ballantine: 379-398.

Bednarik, R. (1995). "Concept-mediated marking in the Lower Palaeolithic." Current Anthropology 36: 605-634.

Bermudez de Castro, J. M., J. L. Arsuaga, et al. (1997). "A hominid from the Lower Pleistocene of Atapuerca, Spain: Possible ancestor to Neandertals and modern humans." Science 276: 1392-1395.

Binford, L. R. (1985). "Human ancestors: Changing views of their behavior." Journal of Anthropological Archaeology 4: 292-327.

Bock, G. and G. Cardew, Eds. (1997). Characterizing Human Psychological Adaptations. Ciba Foundation Symposia. New York, John Wiley & Sons.

Boysen, S., G. Berntson, et al. (1987). "Simian scribbles: A reappraisal of drawing in the chimpanzee (Pan troglodytes)." Journal of Comparative Psychology 101: 82-89.

Brain, C. K. and A. Sillen (1988). "Evidence from the Swartkrans cave for the earliest use of fire." Nature 336: 464-466.

Cachel, S. and J. Harris (1995). Ranging patterns, land-use and subsistence in Homo erectus from the perspective of evolutionary biology. Evolution and Ecology of Homo erectus. J. Bower and S. Sartono. Leiden, Pithecanthropus Centennial Foundation: 51-66.

Calvin, W. (1993). The unitary hypothesis: a common neural circuitry for novel manipulations, language, plan-ahead, and throwing? Tools, Language, and Cognition in Human Evolution. K. Gibson and T. Ingold. Cambridge, Cambridge University Press: 230-250.

Clark, G. (1977). World Prehistory. Cambridge, Cambridge University Press.

Cremaschi, M. and C. Peretto (1988). "Les sols d'habitat du site paleolithique d'Isernia la Pineta (Molise, Italie Cantrale)." L'Anthropologie 92(4): 1017-1040.

Dabbs, J. M. J., E.-L. Chang, et al. (1998). "Spatial ability, navigation strategy, and geographic knowledge among men and women." Evolution and Human Behavior 19: 89-98.

Eals, M. and I. Silverman (1994). "The hunter-gatherer theory of spatial sex differences: Proximate factors mediating the female advantage in recall of object arrays." Ethology and Sociobiology 15: 95-105.

Gamble, C. (1994). Timewalkers: The Prehistory of Global Colonization. Cambridge, Mass., Harvard University Press.

Gamble, C. (1999). The Palaeolithic Societies of Europe. Cambridge, Cambridge University Press.

Gangestad, S. (1997). Evolutionary psychology and genetic variation: non-adaptive, fitness-related, and adaptive. Characterizing Human Psychological Adaptations. G. Bock and G. Cardew. New York, John Wiley and Sons: 212-230.

Gaulin, S. and H. Hoffman (1988). Evolution and development of sex differences in spatial ability. Human Reproductive Behavior: A Darwinian Perspective. L. Betzig, M. Borgerhoff Mulder and P. Turke. Cambridge, Cambridge University Press: 129-152.

Harris, J. W. K. (1983). "Cultural beginnings: Plio-Pleistocene archaeological occurrences from the Afar, Ethiopia." African Archaeological Review 1: 3-31.

Howell, F. C., M. Kleindienst, et al. (1972). "Uranium series dating of bone from the Isimila prehistoric site, Tanzania." Nature 237: 51-52.

Isaac, G. (1984). The archaeology of human origins: studies of the lower Pleistocene in East Africa 1971-1981. Advances in World Archaeology. F. Wendorf and A. Close. New York, Academic Press: 1-87.

James, S. (1989). "Hominid use of fire in the Lower and Middle Pleistocene: A review of the evidence." Current Anthropology 30(1): 1-26.

Jones, P. (1981). Experimental implement manufacture and use: a case study from Olduvai Gorge, Tanzania. The Emergence of Man. J. Young, E. Jope and K. Oakley. London, The Royal Society and the British Academy: 189-195.

Klein, R. (2000). "Archeology and the evolution of human behavior." Evolutionary Anthropology 9(1): 17-36.

Kohn, M. and S. Mithen (1999). "Handaxes: products of sexual selection?" Antiquity 73: 518-26.

Kosslyn, S. (1994). Image and Brain: The Resolution of the Imagery Debate. Cambridge, Mass., MIT Press.

Kosslyn, S., W. Thompson, et al. (1998). "Neural systems that encode categorical vs. coordinate spatial relations: PET investigations." Psychobiology 26(4): 333-347.

Kosslyn, S. M. (1999). "If neuroimaging is the answer, what is the question?" Philosophical Transactions of the Royal Society of London, B 354: 1283-1294.

Linn, M. C. and A. C. Petersen (1986). A meta-analysis of gender differences in spatial ability: Implications for mathematics and science achievement. The Psychology of Gender. J. Hyde and M. Linn. Baltimore, Johns Hopkins University Press: 67-101.

Lumley, H. d. and Y. Boone (1976). Les structures d'habitat au Paleolithique inferieur. La Prehistoire Francaise. H. d. Lumle. Paris, Centre National de la Recherche Scientifique. 1: 625-643.

Marshall, J. C. and G. R. Fink (2001). "Spatial cognition: Where we were and where we are." Neuroimage 14: S2-S7.

McGrew, W. C. (1989). "Comment on "Hominid use fire in the Lower and Middle Pleistocene" by Steven James." Current Anthropology 30(1): 16-17.

Mithen, S. (1996). The Prehistory of Mind. London, Thames and Hudson.

Moffat, S. D., E. Hampson, et al. (1998). "Navigation in a "virtual" maze: Sex differences and correlation with psychometric measures of spatial ability in humans." Evolution and Human Behavior 19: 73-87.

Morris, D. (1962). The Biology of Art. London, Methuen.

Noble, W. and I. Davidson (1996). Human Evolution, Language, and Mind: A Psychological and Archaeological Inquiry. Cambridge, Cambridge University Press.

Oakley, K., P. Andrews, et al. (1977). "A reappraisal of the Clacton spear." Proceedings of the Prehistoric Society 43: 1-12.

Piaget, J. and B. Inhelder (1967). The Child's Conception of Space. New York, Norton.

Potts, R. (1988). Early Hominid Activities at Olduvai. New York, Aldine.

Rightmire, G. P. (1992). "Homo erectus: Ancestor or evolutionary side branch?" Evolutionary Anthropology 1(2): 43-49.

Rightmire, G. P. (1998). "Human Pleistocene in the Middle Pleistocene: The role of Homo heidelbergensis." Evolutionary Anthropology 6(6): 218-227.

Roberts, M. A. and S. A. Parfitt (1999). Boxgrove: A Middle Pleistocene hominid site at Eartham Quarry, Boxgrove, West Sussex. London, English Heritage.

Roberts, M. B., C. B. Stringer, et al. (1994). "A hominid tibia from Middle Pleistocene sediments at Boxgrove, UK." Nature 369: 311-313.

Robson Brown, K. (1993). "An alternative approach to cognition in the Lower Palaeolithic: The modular view." Cambridge Archaeological Journal 3: 231-245.

Roebroeks, W., N. J. Conard, et al. (1992). "Dense forests, cold steppes, and the Palaeolithic settlement of northern Europe." Current Anthropology 33(5): 551-586.

Rohles, F. and J. Devine (1967). "Further studies of the middleness concept with the chimpanzee." Animal Behavior 15: 107-112.

Schick, K. and N. Toth (1993). Making Silent Stones Speak: Human Evolution and the Dawn of Technology. New York, Simon & Schuster.

Schick, K., N. Toth, et al. (1999). "Continuing investigations into the stone tool-making capabilities of a Bonobo (Pan paniscus)." Journal of Archaeological Science 26: 821-832.

Schiller, P. (1951). "Figural preferences in the drawings of chimpanzee." Journal of Comparative and Physiological Psychology 44: 101-110.

Schlanger, N. (1996). "Understanding Levallois: Lithic technology and cognitive archaeology." Cambridge Archaeological Journal 6(2): 231-254.

Silverman, I., J. Choi, et al. (2000). "Evolved mechanisms underlying wayfinding: further studies on the hunter-gatherer theory of spatial sex differences." Evolution and Human Behavior 21: 201-213.

Singer, R., B. G. Gladfelter, et al. (1993). The Lower Paleolithic Site at Hoxne, England. Chicago, University of Chicago Press.

Smith, D. (1973). "Systematic study of chimpanzee drawing." Journal of Comparative and Physiological Psychology 82: 406-414.

Stout, D., N. Toth, et al. (2000). "Stone tool-making and brain activation: Positron emission tomography (PET) studies." Journal of Archaeological Science 27: 1215-1223.

Tattersall, T. (2000). "Paleoanthropology: The last half-century." Evolutionary Anthropology 9(1): 2-16.

Thieme, H. (1997). "Lower Palaeolithic hunting spears from Germany." Nature 385: 807-810.

Toth, N. (1985). "Archaeological evidence for preferential right-handedness in the Lower and Middle Pleistocene, and its possible implications." Journal of Human Evolution 14: 607-614.

Toth, N. (1985). "The Oldowan reassesed: A closer look at early stone artifacts." Journal of Archaeological Science 12: 101-120.

Toth, N. and K. Schick (1986). The first million years: the archaeology of protohuman culture. Advances in Archaeological Method and Theory. M. Schiffer. New York, Academic Press. 9: 1-96.

Toth, N., K. Schick, et al. (1993). "Pan the tool-maker: Investigations into the stone tool-making and tool-using capabilties of a bonobo (Pan paniscus)." Journal of Archaeological Science 20: 81-91.

Turnbaugh, W. A., R. Jurmain, et al. (1999). Understanding Physical Anthropology and Archaeology. Belmont, CA, West.

Tyler, C. W. (1996). Human symmetry perception. Human Symmetry Perception and its Computational Analysis. C. W. Tyler. Utrecht, VSP: 3-21.

Ungerleider, L. G. (1995). "Functional brain imaging studies of cortical mechanisms for memory." Science 270: 769-755.

Villa, P. (1983). Terra Amata and the Middle Pleistocene archaeological record of southern France. Berkeley, University of California Press.

Wagemans, J. (1996). Detection of visual symmetries. Human Symmetry Perception and its Computational Analysis. C. W. Tyler. Utrecht, VSP: 25-48.

Walker, A. and R. Leakey, Eds. (1993). The Nariokotome Homo erectus skeleton. Cambridge, Harvard University Press.

Washburn, D. K. and D. W. Crowe (1988). Symmetries of Culture: Theory and Practice of Plane Pattern Analysis. Seattle, University of Washington Press.

Wenban-Smith, F. (1989). "The use of canonical variates for determination of biface manufacturing technology at Boxgrove Lower Palaeolithic site and the behavioral implications of this technology." Journal of Archaeological Science 16: 17-26.

Whittaker, J. C. and G. McCall (2001). "Handaxe-hurling hominids: An unlikely story." Current Anthropology 42(4): 566-572.

Wright, R. (1972). "Imitative learning of flaked tool technology: The case of an orangutan." Mankind 8: 296-306.

Wynn, T. (1979). "The intelligence of later Acheulean hominids." Man 14: 371-391.

Wynn, T. (1981). "The intelligence of Oldowan hominids." Journal of Human Evolution 10: 529-541.

Wynn, T. (1985). "Piaget, stone tools, and the evolution of human intelligence." World Archaeology 17: 32-43.

Wynn, T. (1989). The Evolution of Spatial Competence. Urbana, University of Illinois Press.

Wynn, T. (1993). "Two developments in the mind of early Homo." Journal of Anthropological Archaeology 12: 299-322.

Wynn, T. (1998). "Did Homo erectus speak?" Cambridge Archaeological Journal 8(1): 78-81.

Wynn, T. and W. C. McGrew (1989). "An ape's view of the Oldowan." Man 24: 283-298.

Wynn, T. and F. Tierson (1990). "Regional comparison of the shapes of later Acheulean handaxes." American Anthropologist 92: 73-84.

Wynn, T., F. Tierson, et al. (1996). "Evolution of sex differences in spatial cognition." Yearbook of Physical Anthropology 39: 11-42.

Yellen, J., A. Brooks, et al. (1995). "A Middle Stone Age worked bone industry from Katanda, Upper Simliki Valley, Zaire." Science 268: 553-556.