Epistemological Problems in Cognitive Archaeology: an Anti-Relativistic Proposal Towards Methodological Uniformity

doi 10.4436/JASS.91003 JASs Invited Reviews e-pub ahead of print Journal of Anthropological Sciences Vol. 92 (2014), pp. 7-41

Epistemological problems in Cognitive Archaeology: an anti-relativistic proposal towards methodological uniformity

Duilio Garofoli1 & Miriam Noël Haidle1,2

1) Zentrum für Naturwissenschaftliche Archäologie, Eberhard Karls Universität Tübingen, Rümelinstr. 23, 72070 Tübingen, Germany e-mail: [email protected] 2) Heidelberg Academy of Sciences and Humanities, Research Group “The Role of Culture in Early Expansions of Humans,” Research Institute Senckenberg, Senckenberganlage 25, D-60325 Frankfurt/ Main, Germany

Summary - Cognitive archaeology (CA) has an inherent and major problem. The coupling between extinct minds, brains and behaviors cannot be investigated in a laboratory. Without direct testability, there is a risk that theories in CA will remain merely subjective opinions in which “anything goes”. To counter this risk, opponents of relativism originally argued that CA should adopt a method of validation based on “indirectly” testing inferences from the archaeological record. In this paper, we will offer a two-part analysis. In the first part, we will discuss problems with the original anti-relativistic agenda. While we agree with the necessity of developing a rational methodology for this discipline, in our view previous analyses have significant weak points that need to be strengthened. In particular, we will propose that “indirect testability” should be superseded by a methodology based upon deductive mappings from networks of theories, followed by a plausibility-selection stage. This methodology will be implemented by adopting an extension of Barnard´s (2010b) proposals for mapping hierarchical systems. In the second part, we will compare our methods with those currently adopted in the CA debate. From this analysis, it will emerge that some proposals in CA are inconsistent with our methodology and are incommensurable with those that are consistent with it. Furthermore, we will show that theories in CA can advance contradictory conclusions precisely because they have been developed using different methods. We conclude that a universal methodology, like that proposed here, is needed for CA to become more objective. It is also crucial for creating conditions for coherent and productive debate among different schools of thought in the field of cognitive evolution.

Keywords - Cognitive archaeology, Epistemology, Incommensurability, Theoretical Mapping.

Introduction could not be evaluated for how well they account for the target phenomenon on any objective cri- Twenty years ago cognitive archaeology (CA) teria. Indeed, analyses were limited to ad homi- emerged out of the Processual school and sought nem and “inside” methods. In order to be able to distance itself from the subjective/interpreta- to interpret the archaeological record, investi- tive approach adopted by the post-processualists. gators typically relied on their own subjective Interpretations were considered to be explana- experience to situate themselves within the envi- tions open to manipulation that would have sioned physical or social contexts of distant pasts, served to align an analysis of the archaeological (Hodder, 1984, 1986, 1987; Johnson, 1999, record to the authors´ views. The interpretations Chapter 7; Trigger, 2007, Chapter 9-10; Shanks

the JASs is published by the Istituto Italiano di Antropologia www.isita-org.com 8 Cognitive Archaeology

& Tilley, 1987a,b; see also Binford, 1987). The implicitly and less evident than in the past. This pioneers of cognitive archaeology criticized this threatens to impede progress towards the emer- approach as a form of opinion, based only on gence of a rigorous discipline of CA with a uni- personal likes – it was “as wished for” archaeol- fied and coherent community of practice. ogy (Renfrew, 1989, 1994; Bell, 1987, 1991, In this paper we propose a revision to the 1994a). The shared aim of these pioneers was to original anti-relativistic agenda, updating it to produce a methodology, with a clearly defined address the new epistemological challenges that set of rules, to enable assertions about prehistoric have emerged over the twenty-year lifespan of cognition to be systematically tested. CA. Over this lifespan the intellectual landscape The original epistemological objectives grad- has benefited substantially from rich and varied ually faded into the background as new theories, contributions from many disciplines. As we show deeply different in form and content, were devel- later, arguments often navigate a specific course oped to provide explanations about the evolu- through intricate networks of related but quali- tion of mind and the emergence of behavioral tatively different theories. practices considered unique to modern humans. As a fundamental assumption, we shall take Several frameworks, ranging from evolutionary for granted familiarity with the original anti- psychology (Mithen, 1996) through cognitive relativistic agenda (Bell, 1987, 1991, 1994a,b; neuroscience (Coolidge & Wynn, 2005; Wynn Binford, 1987; Renfrew, 1994). Our focus will be et al., 2009; Wynn & Coolidge, 2011) to com- upon selecting points in the context of presenting putational theory (Barnard et al., 2007; Barnard, a novel, comprehensive methodology and space 2010a), were proposed to account for the proper- precludes an extensive discussion of the full range ties of human mind and behavior (see Davidson, of issues associated with subjective interpretations. 2010 for review). These approaches, while Moreover, in the meta-epistemological debate, grounded in different specific conceptions of the a radical argument against emotionally driven mind, nonetheless shared a common concern approaches has been championed by proponents with providing cognitive/biological mechanisms of the objectivist movement (Rand, 1964, 1967; underlying behavioral enhancements. A quite Peikoff, 1981). In particular, these authors claimed distinct tradition, rooted in the archaeological that emotions arise as by-products of the process domain, sought to explain the same enhance- of value-formation. In consequence, they cannot ments purely on the basis of socio-cultural inter- be used to assess the validity of those same val- actions between individuals, without reference to ues. Objectivists reached the conclusion that emo- biological constructs (d´Errico & Stringer, 2011; tions are irrelevant for establishing whether judg- d´Errico, 2003; Hovers & Belfer-Cohen, 2006; ments are true or false. Clearly, if we accept that Zilhao et al., 2010). This archaeological tradition the only means of validation are individual likes uses as evidence for its position the presence of and tastes, then the whole idea of “convincing” behavioral practices commonly associated with people of the validity of one theory over another Upper Paleolithic populations in the artifactual loses any meaningful foundation. Acceptance of record of early modern humans in Africa, as well theories would be reduced to, for example, just as non-modern populations in Europe. counting how many people supported a particu- It is clear that these different theoreti- lar theory´s contents. In contrast, we agree with cal proposals are not simply variations within objectivists’ view that there is simply no need a single school of thought. They range across to surrender to this pessimistic position: reason, many specific strands of argument using dis- indeed, can deliver a reliable method for validat- tinct approaches and methods that emanate ing theories and this represents the general goal of from different communities of practice. Within epistemological research. Our revision of the anti- this broader intellectual landscape, new sources relativistic agenda will be grounded, therefore, on of relativism continue to flourish, perhaps the assumption that, contra the post-processualist D. GarofoliJ. Hawks & M. N. Haidle 9

Fig. 1 - Network of inductive, deductive and explanatory inferences (E.I.) among neural, mental and behavioral systems. school, rational criteria can be provided to vali- of alternative theories. This framework has to use date theories about extinct minds, brains and a theoretical vocabulary of considerable range behaviors. Against this background, we will start and precision and this terminology is compre- by exploring the logical foundations required to hensively specified in the Glossary at the end of establish a methodologically coherent CA. In the paper. particular, we will demonstrate the limitations of approaches that aim to provide explanations for the properties of the mind by drawing direct and How to get from artifacts to a theory unidirectional inferences from the archaeological of mind? record. Instead, we will discuss the need for CA to embrace a deductive perspective, which can allow The formulation of a universal method for CA networks of theories, constructed and mapped is an ambitious and intricate problem to address. between multidisciplinary domains, to provide It implies a fundamental premise about the logi- explanations of extinct minds´ properties. Since cal operations that need to be adopted to estab- we cannot bring extinct minds into a laboratory, lish coherent connections between extinct brains, we will suggest that the concept of empirical test- minds and artifacts in the record. An approach ability introduced by Bell (1994a) is insufficient. that aims to account for this problem must nec- For example, we will later show that symbolic essarily deal with multiple sources of data and thought cannot be inferred from the simple pres- types of theory. These need to be mapped one to ence of beads or pigments in the archaeological another in manner that supports justified infer- record. Such evidence needs to be augmented ences. Neural, mental and behavioral systems are with criteria of plausibility and logical validation inter-related entities that exert reciprocal explana- to create an efficient strategy for selecting more tory influences one upon another. Although other viable theories from among less viable ones. system levels, such as social-cultural or bodily sys- With foundational rules for a CA methodol- tems, are clearly relevant for CA, we will focus ogy in place, we will then explore the epistemo- on just three levels to illustrate key points in our logical problems that can potentially confound argument. Figure 1 shows the network of induc- meaningful comparisons of theories. Case studies tive, deductive and explanatory inferences link- will be used to demonstrate the actual existence ing interdisciplinary data and theory within and of these problems in current key debates within among these three qualitative distinct systems. CA. Specifically, we will elaborate how our pro- So, for example, research in neuroimaging, posed methodological framework provides the lesion studies, psychophysiology and neurobiol- conditions and deep structure for what should ogy using extant species enables the induction of hold for the proper and meaningful comparison theories of neural systems. Similarly, experimental

www.isita-org.com 10 Cognitive Archaeology

Fig. 2 - Problematic inductive inferences within and between systems. T research in the field of cognitive psychology and mapped deductively on the past and can be tested. comparative cognition, for instance, can be used Handaxes have been argued to fulfill many to build increasingly more abstract theories of functions, for example, in butchery (Mitchell, how the mind works, while data from cultural 1995), sexual selection (Kohn & Mithen, 1999), anthropology, ethnography and ethology can do as weapons (Samson, 2006) or merely as by- the same with a theory of behavior. At the same products of manufacturing flakes (Davidson & time, the suite of theories makes it possible to Noble, 1993). Handaxes could be described as draw deductive inferences that allow new data to butchery tools because they have some necessary be categorized and explained. and sufficient properties that allow us to include While most researchers would acknowledge them in a theoretical category that defines how that a logical framework of this type is required a butchery tool ought to be. Inferences about within CA, the literature itself is replete with how extinct species behaved with them require examples of inferences that short-circuit the theories about the value of particular properties requirements it implies. Some classes of problem- such as sharp edges, the presence or absence of atic inferences are highlighted schematically in wear, symmetry, the practicability of handling Figure 2. Dashed lines represent inductive infer- them safely or the contexts in which they were ences that should not be made from archaeologi- uncovered. Theories provide the necessary scaf- cal data to minds and brains, while the embold- folding for inferences and for these examples the ened arrow highlights inappropriate inductions properties in focus pertain only to behavior, we from archaeological data to behavioral theory. The have deliberately made no reference to properties overall organization here is crucial. Archaeology is of mental or neural systems. This point applies limited to the behavioral domain and this pro- only to inductions within behavioral systems; hibits direct inductive inferences. Given that cer- the problems are more acute in terms of what is tain forms of inductive inference are problematic required to move among system levels and this within a single behavioral system without making problem will be addressed later. inappropriate assumptions, then the problems It will be recalled from the previous section compound when seeking to make inductive infer- that Bell’s candidate solution to the problem ences from one system level to another. posed by the post-processualists was to provide a The properties of artifacts in the record can universal method for CA that would indeed ena- only fully be explained “by means of” a theory ble the validity of claims about prehistoric minds of behavior, which is built in the present and to be directly tested on the artifactual record. It is D. GarofoliJ. Hawks & M. N. Haidle 11

Fig. 3 - Schematic description of the structure and functioning of a methodology for CA proposed in this paper. clear that Bell thought that he could empirically represents the starting point for testing explana- validate properties of minds from the record. In tions about the properties of the extinct mind. doing so he conflated theories of behavior with Some scholars have routinely drawn inferences theories of the mind. In the next section we will from isolated parts of the artifactual record to illustrate this form of conflation in detail for the specific characteristics of the mind, adopting a case of artifacts for measuring weight and infer- notion of empirical testability, where, for exam- ences about mental constructs. It should be clear, ple, the presence of flower pollen in a grave is from the connective links in Figure 2, that empir- taken to be indicative not only of ritual behavior ical testability within archaeology is limited to the but a mind able to process symbols. To overcome confirmation of behavioral hypotheses that are these two problems we will introduce a holistic deductively constructed from theory of behavior perspective for inductive, deductive and explana- and can be tested in the archaeological record. No tory inference. Networks of inter-related theories empirical proof can be provided to infer directly of cognition, behavior and neural systems are from artifacts to the properties of the mind. Even mapped onto the artifactual record in order to if, as Bell implicitly argued, the behavioral system explain properties of extinct minds and brains is constructed by a proper deduction from theory that cannot be tested through explicit behavioral of behavior and tested empirically on the artifac- experimentation. tual record, properties of behavior are not suffi- Figure 3 focuses on mental and behavioral cient alone to explain the properties of the mind systems to indicate the general structure and connected with it. Even the more solid empirical functional rules for the method that we are going support of statements in the artifactual record to propose for CA. This summarizes the argu- cannot be used to test the characteristics of the ments we have already introduced while also extinct minds connected to them. anticipating the content of what follows. In con- It is well known that a given pattern of trast to Figures 1 and 2, this new diagram high- behavior can be open to explanation by alterna- lights three features. First, it makes prominent tive theories, and this is obviously true for extinct the role of deduction from theories to proper- as well as extant minds. Theory necessarily ties of minds and of artifacts. Second, within this

www.isita-org.com 12 Cognitive Archaeology

bigger picture it is important to note that a key b) existence of units <-> existence of modular role is preserved for induction and deduction measure within research on mental and behavioral sys- c) hierarchy of units <-> existence of a system of tems. Third, it shows there is no direct connec- numeration tion between archaeological data or properties of artifacts to properties of extinct minds. Rather, In other words, the discovery of calibrated the connection is mediated indirectly from prop- stones suggests both the existence of a concept erties of artifacts via theories of behavior through of weight and of constituent units (a), which are theories of minds to specific properties of minds. necessary to support the idea of modular meas- This schema alone does not cleanly resolve ure, (b). Moreover, the existence of a system of the problem of providing objective criteria for numeration, organized in numerical categories, empirical validation of theories – there is no follows from the assessment of the hierarchical escape from the problem of not being able to relationship among units. Therefore, the asser- directly test explanations of mental systems that tion of calibration is directly testable on the arti- no longer exist. However, a proposal for non- factual record, by seeking for similar stones that empirical validation of theories will be advanced hold no weight relations and can therefore invali- that arguably provides a rigorous framework for date the inference in (a). Conversely, the asser- selection that is sufficient to overcome relativism. tions (b) and (c), are not refutable from a direct observation of the archaeological evidence (to Problem of testability this goal there ought to be written material from In the second chapter of the seminal book this society that documents the use of system of “The Ancient Mind”, James A. Bell (1994a) numeration, for instance), but they can be in any argued that archaeologists should aim to con- case rejected by considering their logical connec- struct testable theories of prehistoric cognition tion to (a), which is both testable and necessary as opposed to interpretations. According to his to support the validity of (b) and (c). view, theories must be constituted by statements This indirect approach to testability is for- emerging directly from observation of the arti- malized by adopting the concept of entailment, factual record and that, in turn, can be empiri- which Bell (1994a, p. 19) refers to with the cally tested on it. At the same time, other state- expression: ments might be derived, which are not directly connected to the artifactual record, but can be “if X entails Y, then a mistake in Y indicates that subject to empirical refutation by virtue of logi- there is a mistake in X”. cal connections with assertions that can be conversely tested on the archaeological evidence. Therefore, in order to produce an entirely To help understand this logic, Bell revised the testable theory, statements not directly testable famous Renfrew´s (1982) example of the stone- themselves must entail statements which are based system of weights from the four thousand directly testable, so that the absence of clear evi- year old site of Mohenjo-Daro, in the Indus val- dence of a system of units in Renfrew´s example ley. By finding evidence of the existence of cubic necessarily would lead to the invalidation of any stones, whose weights are multiples of the same potential conclusion concerning the existence of unit, Renfrew advanced a theory of the cognitive numerical systems. Indeed, the logic of entail- processing of weight systems in this civilization. ment implies the obvious rationale that there is A relevant part of his argument can be structured no concept of weight without the identification in the following way (where <-> = implies): of artifacts that can fulfill the role of units, as well as no system of numeration. a) discovery of calibrated stones <-> concept of In the previous section, we discussed the dif- weight + existence of units ficulties associated with empirical testing and D. GarofoliJ. Hawks & M. N. Haidle 13

inference (Figs. 2, 3). In the case of Renfrew´s This general point was originally discussed in example of the Mohenjo-Daro weight system, the the context of Wynn´s seminal review of cogni- argument is constructed through a proper deductive evolution (2002). In his comments on that tion from theory of behavior, which enables us review, Deregowski argued that rotation of tri- to understand how a system of weights should dimensional figures and estimation of symmetry be universally constituted. However, the funda- were not necessary to produce Early Acheulean mental problem here lies in the deceptive convic- handaxes. An alternative, but easier strategy tion that statements about prehistoric cognition can give the same result. In this case, a simple are being tested, whilst actually testing assump- mechanism of perceptual priming would auto- tions applying within the behavioral system. matically have led to choosing the correct shape As seems to be the case with much of the first when presented with two possible alternatives. wave of CA (see Renfrew, 1994; Preucel, 2007), The hard epistemological challenge for CA is both Renfrew and Bell focused on relationships to put the flesh on the mere bones provided in that connect several components of a behavioral Figure 3 to answer the question “how is it pos- pattern, namely the stones and their potentially sible to test assertions about the evolution of the calibrated weights, in order to categorize the use human mind from the archaeological record?” of units of measure in a system of numeration. In framing an answer to this question, it is neces- Classifying the use of calibrated stones with the sary to deal with the post-processualist counter- label “system of weight” tells us little about the arguments to Bell’s agenda. nature of the mental processes required to use it Given what has been argued so far, CA must and the concept of weight that stems from this chart a new and clear epistemology to avoid the analysis can be different from the one we cur- black and white choice of two wholly unpro- rently hold. Indeed, when referring to the sin- ductive options. One option remains Lewis gle behavioral system, testability is warranted by Bindford’s (1987) materialism, according to direct inference from the archaeological record or, which drawing psychological inferences from as illustrated above, through the logic of entail- material facts, via “paleopsychology” (Binford, ment. It is not by chance that Bell more than once 1965), is of dubious value. The other option is used the expression “indirect way for empirical relativism, where anything goes, according to the refutation of statements” (our emphasis). On the authors´ interpretations, likes and so forth. In other hand, when mental processes are considered, the next section we will explore a new epistemol- the shift from a behavioral system to a mental one ogy offering some precise conceptual scaffolding cannot be informed by empirical analysis alone, for the “logical”, rather than empirical, valida- either directly or by entailment. For example, it tion of explanations. is possible to characterize the behavioral practices in different hominid species with complex logical maps (Haidle, 2009, 2010, 2012; Lombard & The deductive approach Haidle 2012). However, here the behaviorist fal- lacy holds. You cannot infer the identity of under- Structure of the deductive method lying cognitive processes from observation of the Our proposal to develop a new epistemol- behavior alone, even with an abstract schema of ogy focuses on use of a deductive framework to the sort used in our illustration of a system of explain the archaeological record. This deduc- numeration. Therefore, specific behavioral prac- tive method aims to map a suite of theories tices, considered in isolation from the rest of the that account for systems of interest and how behavioral architecture that characterizes one spe- they behave. Behavioral, mental and neural sys- cies/population, might be sustained by a differ- tems are causally inter-related, each with their ent pattern of cognitive operations and ultimately own qualitatively distinct architectural proper- associated with different mental capabilities. ties, and exert reciprocal constraints one upon

www.isita-org.com 14 Cognitive Archaeology

Fig. 4 - Three-dimensional representation of the first stage of our deductive method proposed for CA. The spheres represent the three assembled systems (N=neural, M=mental, B= behavioral), mapped horizontally through explanatory inference, in order to form meta-theories of system-interaction. another. As we explain later, in order to better rather than piecemeal way. To realize our method, understand these reciprocal influences, we need we have to organize the deductive framework and a well defined set of macro-and micro-theories of specify the first major premise that can lead to how each type of system works (Barnard et al., general principles of interaction that govern the 2000). The notions of macro and micro-theory behavior of any system. Figure 4 illustrates this are deeply rooted in a hierarchical conception by referencing two new constructs: Meta-theories of how assembled systems are constituted and that relate system levels for a given species (e.g. serve to explain how entities that are part of each Homo sapiens sapiens and Pan paniscus), and the system behave (Newell, 2000). A macro-theory concept of Core-Theory which aims to capture can be roughly defined, in fact, as a theory that what governs interactions among system levels explains how the subcomponents of an assem- for all species across any and all system states. The bled system interact, while each of these sub- later construct and its detailed properties will be components requires in turn a micro-theory to incrementally built up as our argument devel- explain the properties of its constituent parts. ops. For the present, it is sufficient to note what The challenge is further complicated by the Figure 4 highlights is that Core-Theory must fact that our focus in CA is on understanding be induced from Meta-Theories. In this paper, the neural, mental, and behavioral systems of Figure 1 diagrammatically introduces the func- not just one species of hominoids, but many. tion of an explanatory inference and its more pre- Because multiple macro- and micro-theories of cise definition can now be stated. “Explanatory qualitatively different systems and many species inference” is the logical operation adopted to are required, we clearly need a way of thinking construct a particular Meta-Theory and it refers about all these inter-relationships in a systematic to the bidirectional explanatory power these D. GarofoliJ. Hawks & M. N. Haidle 15

Fig. 5 - Vertical mapping from the core-level allows to complete the missing fields (interrogative marks) in the target level, in relation to the structure of the behavioral architecture in the target level. systems reciprocally hold. This procedure of hori- is required to infer properties of the missing tar- zontal mapping is repeated for different species/ get mental and neural systems. This method, as populations and many meta-theories are pro- depicted in the figure, can be represented with duced until all of these are aggregated in a syn- the following expression in formal logic: thetic “core-theory”. The value of core-theory is that it should ena- If X<->Y<->Z ble us to predict through explanatory inference, and X = X* for instance, how the variation in the architec- then X*<->Y*<->Z* ture of a mental system (M) affects a behavioral system (B), by virtue of the principles that uni- In line with the structure of deductive argu- versally bridge and constrain the various systems. ments in logic, the validity of the conclusion fol- However, a core-theory appears as a pure abstrac- lows directly from true premises and for every tion and should be considered as a way to account given X it would be possible to find a specific Y for the comparisons of meta-theories and the and Z, by virtue of the universal rules that are extraction of invariant rules of system behavior. implicitly stated in the major premise. The next step for CA is to recruit the power Figure 5 is a refinement of our earlier Figure 3 of deduction to perform a vertical mapping from that now makes clear the role of explanatory infer- the level of a core-theory to a target level of inter- ence. If a direct horizontal mapping is impossi- est (Fig. 5). In CA, only one system is available ble from the target behavioral architecture to the within the target level - the behavioral architec- other systems in the same level, the only way to fill ture found in the artifactual record, albeit where the gaps of knowledge results in vertically infer- the evidence of the complete system is necessar- ring them by virtue of the universal principles of ily partial. The obvious point is that deduction connection that bind elements in the accessible

www.isita-org.com 16 Cognitive Archaeology

al., 2007). According to (a), increasing the capacity of the neural system, for instance by alter- ing the architecture of constituent networks, increases also the capacity of the cognitive one. Therefore, the behavioral repertoire that can be potentially handled by these enhanced systems is also “increased”. At the same time, our second assumption (b) implies that those principles applied in the same way in the ancient past as in the present. It is well known, as Hume (1739) argued, regularities in the present will not necessarily apply in the future, but we believe we can trust that biological principles past and present are congruent simply by virtue of the evolution- Fig. 6 - Two-dimensional representation of the proposed methodology (implemented theory). ary relations that link organisms. Our proposed methodology shares with Bell’s original conception the property of assuming indirect strategies as a tool to validate theories. The main difference meta-theories and are synthesized by the core- between our approaches, however, lies in the fact theory. In other words, the diagram, as formu- that we do not recognize the logic of entailment lated, tells us that having different perspectives on as sufficient to check the validity of statements how neurons, cognitive systems and hominoids between two different aggregated systems. behave, warrants sufficient generality to infer As an illustration of the difference between from them how a given behavioral architecture, using our holistic and deductive approach as detected in the artifactual record and linked to opposed to that relying on entailment, we can a target species/population, is associated with a consider Acheulean tools. These are created by mental and a neural architecture. “Implemented the staged removal of flakes from a core to create theory” is the way we define the result of the ver- a bifacial and symmetrical entity. Wynn (2002) tical mapping between the core-theory and the argued that mental rotation of three-dimensional behavioral architecture in the target level. In other figures played a key role in the process of manu- words, an implemented theory is the final body of facturing these tools. However, nothing in the theories that is provided in CA to explain proper- record alone can directly be used to argue against ties of extinct minds and brains. A schematic syn- the involvement of mental rotation without a the- thesis of the whole logical process that defines our oretical context to support an appropriate deduc- proposed methodology is shown in Figure 6. tion. In contrast, implications can be invalidated This deductive method obviously relies on in cases such as Renfrew´s example, when stones two assumptions. First, relationships between with non-standardized weights are found. The system levels are subject to deep abstract prin- point is that inferences from evidence across dif- ciples (a), which exist independently from the ferent domains, when isolated from a theoretical specific theories adopted to explain them. This context, can lead to the formulation of distorted means that variations in the properties of a sys- conclusions. In fact, statement X “there are sym- tem necessarily produce changes in related prop- metric relationships between parts of this artifact” erties of connected systems. For example, capacity entails statement Y “symmetry is produced by tri- is a property of both the neural and the mental dimensional rotation of figures in working mem- system that accounts for the amount of infor- ory”, but there is no empirical evidence that can mation a neural network, as well as a cognitive disprove this entailment in the artifactual record, architecture, can potentially handle (Halford et because the mind does not fossilize. D. GarofoliJ. Hawks & M. N. Haidle 17

Problems like these need to be solved only between technological complexity and popula- by checking if the whole behavioral architecture tion size. In particular, Kline & Boyd (2010) of the population that produced those artifacts meta-analyzed a set of ethnographic data on is compatible with a hypothesized mental archi- artifacts and behavioral practices of populations tecture. This needs to be derived via a deduc- living in different Oceania islands and concluded tion from the core-theory that is built upon that islands with small populations had a sim- the firmer foundation of empirically warranted pler marine foraging technology (but see Read, meta-level theories. We need to go beyond the 2012 for a counterargument). More widely, an idea of testing single statements and the adop- extension of the three-systems logic would offer tion of linear chains of inferences characteristic additional constraints and allow a more informa- of Bell´s falsificationist agenda. Instead, we need tive mapping to the archaeological evidence (i.e.: to shift to an alternative conception of “testabil- both theories of the body and demography are ity”, more similar to a form of holistic evalua- informed by paleoanthropology). tion of theories (Godfrey-Smith, 2003, p. 31). In a similar vein, constraints on neural archi- To reconstruct missing mental and neural sys- tecture can come from paleoneurological stud- tems we need to examine not single statements ies on endocasts and neontological enquiries that are entailed, but entire meta-theories. For characterizing relationships between variation a given target behavioral architecture, a core- of brain size, shape and functions in human theory of system behavior would predict proper- and non-human primates. The paleoneurologist ties of mental and neural architectures (i.e.: the Emiliano Bruner, for instance, has demonstrated core-theory that explains the connection among that modern humans, but not highly encephal- systems in modern humans and apes entails the ized Neanderthals, are characterized by a non- conclusions about the mental architectures of allometric expansion of the upper parietal regions other extinct hominoids). of the endocast (Bruner, 2003, 2004, 2008, 2010, In framing our approach so far we have con- 2011), an autapomorphic trait that is supported centrated on only three systems. This should by a specific ontogenetic phase in modern humans not be taken to exclude contributions from that has no counterpart in species that predate macro-theories of other relevant but qualitatively anatomically modern humans (Gunz et al., 2010; distinct systems. Clearly, properties of the bio- Neubauer et al., 2010). This morphological varia- mechanical and sensory systems of a particular tion, which determines the characteristic globular body morphology also constrain relationships to form of the modern human brain (Lieberman et cognitive and neural systems. Such constraints al., 2002), contrasts with the elongated structures may be informed by models of embodied cog- of more archaic populations. This difference has nition (Lakoff & Johnson, 1999; Clark, 2008) been associated with potential disproportional and their application to the study of material cul- white matter expansion and enhancement of ture (Malafouris, 2008a,b,c, 2010a,b; Mithen & parieto-frontal connectivity (Bruner, 2003, 2004; Parsons, 2008). The same applies for constraints Coolidge & Wynn, 2008). This enhanced con- that propagate from higher order social systems. nectivity, in turn, could be a candidate biologi- An account of demographic changes would give cal substrate for selective cognitive advantages more power to deductive mapping, both in the (Jung & Haier, 2007) in modern humans. This horizontal and vertical component. One example hypothesis has been recently tested in neontologi- of this could be the hypothesized link between cal studies on modern humans. A slight correla- group size and rates of cultural innovation, tion between brain globularity and information which comes with at least some authority from processing speed has been demonstrated (Bruner mathematical modeling (Shennan, 2001; Powell et al., 2011). Additionally, these researchers et al., 2009). Furthermore, recent empirical anal- claimed that this effect might have been more yses have proved the existence of a correlation pronounced on an evolutionary time-scale. This

www.isita-org.com 18 Cognitive Archaeology

form of analysis well illustrates how empirically be represented in whole or in part. Even though supported constraints relating system levels can there are candidate unified theories of cogni- contribute to the substance of core-theory. tion (e.g. Newell, 1990) and macro-theories of One key question remains to be addressed. mind (e.g. see Barnard et al., 2000), there are Under circumstances where there is no direct vast numbers of micro-theories to account for empirical evidence to refute theory, and a num- specific phenomena and not much in the way of ber of implemented theories are available, how consensus about how best to decompose minds do we select the candidate that best accounts for or about how individual micro-theories can be the context of the record? reconciled as parts of a bigger picture. However adequate the micro-theories might be within a Problem of validation: criteria for selection discipline, a key challenge for cognitive neuro- Any deductive argument can lead to false con- science, cognitive anthropology and comparative clusions if the premises are false (Godfrey-Smith, psychology would be to develop convincing rules 2003, p. 41). Use of our methodology involves a to connect their own system-level concerns with series of stages (Figs. 5, 6) and false conclusions those at an adjoining level - a task that becomes could arise as a result of the adoption of flawed even harder as the number of constraints to be macro-theories, by performing erroneous hori- accommodated increases. zontal mapping between systems, or by creating A crucial but also hard aspect of the valida- a corrupted core-theory. Even an adequate core- tion process lies with the vertical mapping from theory would allow errors to propagate into the core-theory to the target level “missing” system, vertical mapping, if its minor premise is wrong, be it mental or neural. As argued earlier, evidence as for instance in case where there is some prob- cannot play a part here. The “logical validation” lematic interpretation of the artifactual record, that is performed within our methodological and hence flawed behavioral architectures. In this schema requires criteria. Some criteria for selec- respect, our proposed methodology allows us to tion can be borrowed from David Chalmers´ pinpoint with some clarity where intricate argu- (1995, 1996) analysis of the “hard problem of ments can be flawed. This same property also consciousness” (for a review see Searle, 1997, means that we are going to need two different 2004). This problem shares with CA the funda- stages to decide among candidate implemented mental epistemological issue of inaccessibility of theories. These two stages once again correspond the object of science to scientific methods. The to the two main phases of mapping. subjectivity of first person experience can no more The first stage basically involves the con- be investigated with scientific methods, than can struction of a core-theory. Here, evaluation sim- the minds of extinct hominoids. Chalmers´ solu- ply means that principles must be supported by tion involves a series of non-empirical principles empirical data. Experimental work would serve to evaluate candidate theories of non-observ- to define macro-theories of the assembled sys- able phenomena. The six criteria proposed by tems of interest, by virtue of the study of their Chalmers are adapted below for CA: micro-theoretical components, and to under- a) Internal Coherence: when mapped from a stand the way they constrain each other. This val- core-theory to the target level, the principles idation stage is the “easier” part of the epistemo- that bind architectures must be preserved in logical problem of CA. This should not be taken the target level. In other words, the rules to mean that the construction of a core-theory is adopted to infer properties of the missing an easy task, but that in this phase there is still systems should be coherent with those pre- a connection to hard data. Extracting principles dicted by the core-theory. Not complying of interaction between levels of architecture for with this criterion would imply an incoher- extant species is far from an easy task. There is ent vertical mapping and the whole deduc- not much agreement about how minds should tive process would be corrupted. D. GarofoliJ. Hawks & M. N. Haidle 19

b) Simplicity: Theoretical schema should aim the construction of an implemented theory. Ad for simplicity in any of their component hoc hypotheses, in fact, could be used to resolve parts. In particular, the result of vertical issues with each of the principles derived from mapping should be a simple target level Chalmers (1996) and listed above. They can be meta-theory. However, it is worth noting employed to correct problems of coherence in the that simplicity is not always possible, due to rules adopted to map the various elements within the fact that there are no guarantees that the the whole theoretical schema (crit. a), to cover world is simple (Godfrey-Smith, 2003, p. gaps of knowledge left behind or to deliberately 55). It should be possible to validate com- rule out uncomfortable parts (crit. c-d). They plex theories as well, if the payoff exceeds may also be used to justify contradictory aspects the increase of complexity. of the theory by creating “exceptions” (e), while c) Homogeneity: Theories must show no simplicity (b) results as deeply influenced by the gaps in aspects of reality that they intend number of changes that are produced to align with to cover. the other criteria. It follows that implemented d) Inclusivity: Theories should not deliber- theories that need to be adjusted in any of these ately exclude aspects that cannot be accom- ways are less plausible than those able to perform modated within their logical structure. a coherent mapping without recourse to logical e) Consistency: There should be no contradic- alterations. Therefore, this variant of plausibility tory parts in a single implemented theory. A works by embracing Occam´s razor. Selection statement and its negation cannot be demon- among alternatives is now dictated within an strated within the same theoretical schema. eliminative perspective (Platt, 1964), that rules out all the more implausible alternatives en route David Chalmers, in his book “The Conscious to adopt a preferred theoretical schema. To help Mind” (1996 - Chapter 6), supports the idea that understand this generic logic, we can draw on the six criteria for selection exert constraints on Walton’s (2001) discussion of the case of a juridi- the plausibility of theories and that this can act as cal diatribe in Ancient Greece. Two contenders, a the gold standard for assay in non-empirical situ- big man and a little one, were involved in a fight ations. However, Chalmers does not provide any without witnesses. Each of the contenders had precise definition for this concept. He prefers to to convince the jury that the other had started use an example to discuss the idea that two alter- the fight. It would be implausible for the jury to native theories might be both perfectly rational think that the small man assaulted the bigger man in terms of the logical connections between ele- first, without posing some ad hoc hypothesis to ments, but they can hold a different level of plau- justify this inference. For instance, claiming that sibility. In this way, he imagines the situation of the little man is an expert martial artist would two competing hypotheses. According to the first raise the level of plausibility of the hypothesis that one, the world has been created fifty years ago, he actually started the fight. Martial experience together with all fossils and memories, while in would compensate for lack of size. However, if it the second hypothesis evolution really happened were not possible to empirically check the fact the as we know it. Comparing these hypotheses, one little man was an expert fighter, then there would must conclude that both are rationally conceiv- be no grounds to suppose that this hypothesis able, but the first one is implausible, because it is is actually more plausible than the first one and too complicated and based on a series of unnec- there would be therefore no reason to prefer it. essary assumptions. In CA, reference to evidence is typically more To solve this series of issues, we propose a def- indirect and elusive, so that even the most plausi- inition of plausibility that takes into account the ble implemented theory would quite likely retain number of ad hoc hypotheses that are produced a speculative component in the vertical mapping to connect the core and the target level during phase of the framework. But as we have argued,

www.isita-org.com 20 Cognitive Archaeology

this does not result in epistemic relativism or not universal, but specific to the theory being equivalence in the value of the conclusions: selec- implemented. (i.e.: some parts of one system tion for plausibility is grounded in explicit criteria architecture are mapped onto different parts for theory validation, where Bell´s empirical test- of other architectures, or parts are omitted, ability perspective is not applicable. etc...). Here the mappings are incoherent because their premises are problematic. 2) Incommensurability within architectures: as- Comparing alternative theories sembled systems are mapped correctly, following a universal logic, but the rules of Problem of Incommensurability construction within each mapped system To evaluate the content of alternative imple- are different, so that the system architec- mented theories, it is necessary that the same tures themselves are incongruent. Incon- basic principles are applied while developing gruence, in this case, is the cause of inco- those theories, from micro-theories to the full herent mappings. implementation. Theories of complex systems 3) Incommensurability between architectures: as- are seldom simple and cannot be easily reduced sembled systems are constructed according to unitary elements that can be easily manipu- to the same rules, but the different layers lated. As we shall see, a network of theories at that constitute each architecture (Fig. 7) are different levels may be organized into hierarchi- again incoherently mapped to the layers in cal layers and, when one is mapped to the others, another architecture using a set of different relationships within and among their constitu- rules. (Same system structures, different ent layers must be treated in a disciplined way logic to connect them). (Barnard et al., 2000). If everyone were to use a different set of rules In order to attempt to find a solution to these for mapping among system levels, then the final three problems of incommensurability, we will cognitive archaeological theories that are imple- refer to the model proposed by Philip Barnard mented and compared would be incommensura- and colleagues (2000) for mapping theories, ble. The word “incommensurable” implies that revising it on the light of our methodological the logical structures involved in the construc- framework and suggesting its use as a potential tion of their premises and in making deductions uniform method in CA. differ. Consequently, it would be impossible to properly compare those implemented theories. Barnard´s model: rules and problems Incommensurability can be responsible for con- Barnard et al.’s framework was originally fusion and relativism, because some theories that derived from research on human use of modern are presented as real alternatives to other ones rather than ancient technological systems and are alternatives not because they are theoretically therefore the problems it addressed parallel those advantageous, but because they have been assem- of CA (see Barnard et al., 2000 and reference bled with an improper methodology. At the end therein). A key objective of their work was to cre- of the paper we shall illustrate these problems of ate a method to bridge systemic architectures of a comparison with two case studies of inferences qualitatively different nature (e.g. mental & tech- from the presence of beads found in Neanderthal nological systems) in a coherent way that would and Early Modern settlements. support an understanding of the conjoint behav- Problems of incommensurability can be fur- iors of human and technological interactors. ther refined into three categories: The basic logic of their approach is based on hierarchical decomposition (Fig. 7). Every archi- 1) General Incommensurability: Meta-theories tecture is an assembled system [A], composed of a (see Fig. 4) are assembled with rules that are series of basic interactors [B]s. At the same time, D. GarofoliJ. Hawks & M. N. Haidle 21

Fig. 7 - Hierarchical decomposition of an assembled system into two layers. Reproduced with permission from Barnard (2010b), in Nowell & Davidson: Stone tool and the Evolution of Human Cognition. University Press of Colorado. each of these basic interactors is itself composed differences in mental architecture. Figure 8 now of constituent interactors [C]s. A macro-theory of illustrates a schema for mapping networks of neu- an assembled system is required to explain how ral, mental and behavioral theories. the basic units interact as parts of the overall sys- Assembled systems can be mapped horizon- tem, while micro-theories are required to explain tally by adopting the following basic principles: how the constituents of each basic interactor gov- 1) An assembled system [A] in one layer (e.g.: ern its behavior. So, each interactor is a behavioral here neural or mental) enables its collective entity that occupies a specific position within the capabilities to behave as a basic unit within hierarchy of an assembled system and is so named a system assembled at some superordinate because it interacts with other elements within level, here mental or behavioral (A→B), the same layer. where its properties now constrain the be- The behavior of any interactor is determined havior of the new superordinate assembly. by two main considerations that act concurrently 2) The behavior of a basic unit [B], when incor- on it: (1) the nature of that interactor´s constitu- porated into a superordinate system, must also ents, and (2) the constraints that are exerted on it carry with implications from a relevant body by the other interactors present at the same level of micro-theory from the lower layer (B→C), within the hierarchy. So, for example, according since these also constrain how the relevant B to Barnard (2010b), a behavioral system assem- may behave within the superordinate system. bled to make stone tools might be composed of For example, for system assembled with an a human agent [B1], a hammer [B2] and a core expert stone knapper [B] or an inexpert one [B3]. A tool using system might have the same [B’], the two may share a common macro- human agent [B1] but now replace the core and theory of how their minds are composed the hammer with the tool [B4] and an animal overall, but require different micro-theories of carcass [B5]. Likewise, an hypothesized mental their perceptual, manual and planning skills. architecture of a Neanderthal might be decom- 3) The relationships are bidirectional, in the posed into a particular set of mental subsystems sense that the [B]s in a superordinate sys- (Bs) and that of a modern human into a different tem can constrain an [A] in the subordi- assembly of Bs, with some of these in common nate one (B→A) too (but see below for an between the two architectures and others distinct. alternative conception of the principle of Similarly, their brains would be composed of neu- directionality). To continue with the same ral circuitry with shared cellular electrochemistry example, the particular micro-theory need- but different network architectures whose differ- ed for an expert stone knapper would have ences really can be mapped onto hypothesized arisen only if that individual had a history

www.isita-org.com 22 Cognitive Archaeology

Fig. 8 - Barnard´s diagram for horizontal mapping of the three architectures in cognitive archaeology. Reproduced with permission from Barnard (2010b), in Nowell & Davidson: Stone tool and the Evolution of Human Cognition. University Press of Colorado.

of training and practice in similar behavio- terms from Barnard’s description that the things ral architectures in the past. that interact in neural architecture relate to circuits, cells and their electrochemistry, those that Taken together, it should be clear that proper- interact in mental architecture concern states of, ties and behavior of interactors present in mental and operations on, “information”, while those in and behavioral systems are a product of families of behavioral architecture relate to changes of state constraints and that these tightly connected sys- in animate and inanimate entities in our physi- tems reciprocally influence each other. This idea cal and social worlds. However, for each of these of bidirectional influence can be contrasted with a systems there is a great deal of ambiguity about more neurocentric perspective, where causal rela- what really constitutes a specific interactor and tionships between systems are frequently cast as which level it should be assigned to in a hier- unidirectional and commonly oriented from the archy. This problem is particularly acute when neural to the mental right through to evidence it comes to mental architecture. In the field of in behavioral systems. For example, Klein (1995, cognitive psychology as a whole, there is lit- 2000, 2001, 2008) explained what he considered tle agreement on how we might best define the to be the abrupt emergence of Upper Paleolithic components (i.e.: [B]s) of the mind. Even worse, innovations by reference to a punctiform muta- there is a vast numbers of candidate micro-the- tion in modern human brain architecture, incur- ories applied to specific domains of mental life ring at ca 60 kya. In his view, this neural enhance- among which it is hard to choose on the basis of ment promoted modern human cognitive capa- evidence currently available and little in the way bilities and in particular the extraordinary ability of a body of macro-theory to organize them. We of our species to innovate. have a problem in determining what the [B]s and While Barnard’s framework has much to [C]s are and what layers to assign them to when commend it, there are only a few examples of we apply this approach to CA. its application to practical cases in modern and If the aim of the diagram is to provide a ancient technological systems. It is clear in broad tool that is universally valid, independently of D. GarofoliJ. Hawks & M. N. Haidle 23

the macro-theories we choose to adopt, and the and the endocrine system). The same applies to implemented theories produced with this model behavioral systems. Over the course of a human should be commensurable and accessible to the lifetime and over all the tasks accomplished by criteria for selection we proposed earlier, then it people in different human occupations and roles, is necessary to generate a list of rules of function- there are equally vast numbers of possible behav- ing. These rules should be followed universally ioral systems. Following these assumptions, it by all implemented theoretical proposals. If this appears that all layered architectures cannot be fundamental condition is not met, the model completely inclusive. However, Barnard notes once again risks allowing relativism. In this case (personal communication again) that his focal allowing anything to be an A, a B or a C enables point for theory development is a macro-theory a theorist to propose connections that reinforce of mind that is fully specified, while his strategy their specific theory and make it difficult to com- is to leave adjacent layers only partially specified pare that theory with others. in a manner that most efficiently informs the By way of clarification, though necessary as development and testing of his theory. a starting point, Barnard’s descriptions of the In our view, for the purposes of developing interactors in neural, mental and behavioral this framework for CA, the problem of “optional systems leave us with a problem of ambiguity incompleteness” must be taken into account by to resolve. The main problem with the absence embracing a “realist” agenda that would allow of precise rules for construction of architectures us to recognize the limits of incompleteness. We is that they might be built by adopting differ- should not confuse “optionality” with “anarchy”. ent strategies. As a consequence, arbitrary rules This means that the optional choices in decom- of construction can lead to architectures that posing and configuring architectures must be are incongruent when mapped. It appears from made a priori. They need to be considered in Figure 8 that while mental architecture is defined the stage of epistemological discussion and not a as the entire structure of one species´ mind, con- posteriori, just before the application of a particu- stituted by a set of units that interact, which in lar implemented theory. In other words, the fact turn can be decomposed in constituents, the that architectures cannot be completely inclusive neural architecture does not seem to account for by nature cannot be used as a justification to the entire brain, but only for one extensive part allow any potential manipulation of their struc- of the whole circuitry. The behavioral architecture and hierarchical organization. ture, as well, is focused on just one behavioral Assembled systems need to be as congruent practice and the entities it involves. It does not as possible, starting from the nature of the [A]s, seem to address the entire spectrum of practices which must be set at a comparable level of refer- accounted for by the mind of a particular homi- ence/complexity. Even though we might say that nine and the wider culture within it is embedded. a chosen mental architecture is incomplete, this This problem with the original formula- should not allow us to arbitrarily change the level tion has been recognized by Barnard (personal of reference in the other architectures by replac- communication) and relates to the “Problem of ing the brain with a part of it and then mapping optional incompleteness”. Complex biological sys- the partial neural architecture on the whole men- tems, which involve huge networks of interac- tal one. So, if the [A] in the mental architecture is tions, can hardly be depicted by synthetic dia- intended to represent the overall structure of the grams. This can be seen in neural architecture, mind, then obviously we would perform an inco- for instance, for which a complete description herent mapping if we chose some substructure of the whole set of interactors that constitute such as the arcuate fasciculus´ circuitry to sup- it would require considering almost limitless port this role in underlying neural architecture. amounts of entities and processes (e.g. 1011 neu- Now, examination of Barnard’s dia- rons plus many neurotransmitters, hormones gram reveals a conceptual difference in the

www.isita-org.com 24 Cognitive Archaeology

composition of mental and neural architectures that species may contain evidence that the pres- on the one hand and behavioral ones on the ence of bead ornamentations is open to explana- other, that indicate we need to add to his schema tion by non-symbolic mental capabilities. to achieve full coherence of horizontal mapping. Also, this addition will help counteract inap- Barnard´s model extended propriate inferences from behavioral systems to It is worth exploring in a little more detail mental ones. the kinds of mapping from mental architecture With neural and mental architecture, the set to a behavioral architecture that can lead to dis- of basic units that form the assembly are invari- torted conclusions. Improperly constructing an ant – all “standard” humans have the same num- architecture using a single example or class of ber of mental subsystems/processes and the same behavior, such as ornamentation, rather than a sets of basic neural circuits – although there will, larger “repertoire” of behaviors, represents a vari- of course, be naturally occurring exceptions. ant of the second type of incommensurability, Variation in capability of the system, such as that within architectures. In this case, the same expertise at manual skill, vocal communication macro-theory of mental architecture can be or problem solving, occurs in the level cap- mapped onto different forms of behavioral archi- tured by micro-theory. The same clearly does tecture, resulting in markedly different imple- not apply with behavioral architectures. Across, mented theories. for example, Mousterian tool making, hunting, An example of this class of improper reason- ornamental marking, food gathering and prepa- ing can be illustrated in a thought experiment ration, procreation, caring for the young and old, adapted from work by Wynn (2002). It relates migrating, or conducting simple burial, there is to why incompleteness cannot justify changing significant variation in the Bs that enter into the constituent rules for architectures. Suppose we systems and accordingly variation in both layers were to start producing Acheulean artifacts as of the hierarchy for behavioral systems. a hobby tomorrow. Having made some, we put As expanded upon later, two interconnected them in a secure case where they are found by a points follow from this observation for how we cognitive archaeologist in the future. This arche- should frame use of horizontal mappings. First, ologist could proceed to structure a target level when making horizontal connections from men- where the mental architecture accounts for the tal to behavioral architecture, the selection of entire structure of our mind, while the behavio- what is connected must be made in a manner ral architecture represents the behavioral pattern that is not generic, but sensitive to the target con- underlying stone tool making. To put it bluntly, text. Prior to the invention of writing, stories that the implemented theory that results in this case could not be depicted had to be memorized and by vertical and horizontal mapping could lead to vocally transmitted and learned. Second, there the conclusion that we have a mental architec- is the inverse problem that when attempting to ture comparable to that of the Acheulean popula- make horizontal connections the other way – tions. This is the result of incongruence based on from evidence in a behavioral system to mental the fact that the whole mental architecture (the architecture - a formulation such as Figure 8 may mind) is compared with only a fragment of the be used to justify erroneous generic inferences. behavioral architecture (tool making). However, For example, existence of a basic interactor in a that same future archaeologist would also most single behavioral system involving a target spe- likely find the remains of roads, rocket launch cies, such as bead ornamentations, has been used sites and abstract sculptures forged out of carbon to argue for the generic presence of symbolic fiber. Armed with our method and a well-formed mental capabilities (Zilhao et al., 2010; d´Errico theory, the fuller set of traces of the behavioral et al., 2005; d´Errico & Stirner, 2011). However, systems that made up our material culture would the wider context of other behavioral systems for lead our archeologist to infer that our mental D. GarofoliJ. Hawks & M. N. Haidle 25

Fig. 9 - Illustration of a Poly-Architecture for behavioral systems. Modified with permission from Barnard (2010b), in Nowell & Davidson: Stone tool and the Evolution of Human Cognition. University Press of Colorado. architecture was likely more advanced than those resulting horizontal mapping will be coherent, as of Acheulean populations. Further examples of well as the final stage of vertical mapping for the incommensurability linked with improper con- construction of an implemented theory (Fig. 5). struction of architectures will be discussed in In further developments of this methodology, the later sections. logic of such aggregations could be open to more A solution to incommensurability that can formal representation within the mathematics emerge from the use of incongruent architectures of set theory. If the horizontal mapping and the can capitalize on wider variation in a behavioral construction of the meta-theories are coherent, system that we noted earlier. The many and var- then it means that the call for explanations from ied behavioral practices that are part of a species’ the core-level is correct as well and the mapping repertoire can be used to repeat the mapping pro- can then produce candidate implemented theo- cess and counteract the problem of incomplete- ries, that can be properly submitted to the criteria ness in a single behavioral architecture. The final for selection for plausibility we specified earlier. conclusion from such a sequence of operations would be the same as a proper mapping with Problem of universality congruent architectures. However, in our view Up to this point, our agenda has been this process is not necessary and can be avoided focused on the attempt to define a universal by simply integrating all iterations in a single methodology for CA at the normative level. In poly-architecture. A poly-architecture can be the next sections, we will shift to the descriptive defined as the addition (Σ) of all the individual level in order to create a concrete perception of behavioral practices and systems in a synthetic concepts and mechanisms that have been rather architecture that accounts for the entire behav- abstract until now. In particular, we will compare ioral repertoire (Fig. 9) as well as for the entire the methods adopted in current CA theories brain circuitry (not represented). with the methodology discussed in previous sec- The poly-architecture [A*] is the aggregate of tions. Our goal is to provide examples of theories the partial architectures and the same strategy is that conform to our proposal and theories that used for all the subsystems within the layers of instead do not and therefore need to be adjusted. reference of the same architecture. Each [B*] now As it will become clear from what follows, a stands for all the behavioral processes that are at significant proportion of theoretical outcomes the base of a single practice, while the [C*]s are in CA conform, at least in part, with our pro- the constituent elements that interact to produce posed methodology. Indeed, several theories those processes. This extension of Barnard´s dia- contain parts that are commensurable and of gram now allows us to correctly construct archi- comparable efficiency. Here, we will focus on tectures, so that they will be congruent and the Mithen´s (1996) modular hypothesis, Coolidge

www.isita-org.com 26 Cognitive Archaeology

& Wynn´s (2005) Enhanced Working memory have allowed our species to perform a simultane- and Barnard´s (2010a) Interacting Cognitive ous integration of more complex information. In Subsystems. We will show that while their respec- their view, indeed, a genetic mutation in brain tive macro-theories of cognition differ among the networks at 90-50 kya may have enhanced work- three core-theories, the principles on which these ing memory capacity and, in consequence, the architectures have been assembled are consistent. development of complex tools and behaviors. The proposals are therefore commensurable at An alternative to Coolidge and Wynn´s least within the level of cognitive architecture. model has been recently advanced by Philip At the same time, the behavioral architecture Barnard (2010a) with his Interacting Cognitive depicted by these core-theories seems to be quite Subsystems model (ICS). Within this perspec- invariant, opening to the possibility of a proper tive, evolution of the mind has been described comparison that includes more than one system. as an additive process: cognitive architectures Since there is a paucity of evidence concerning gradually became more complex via the addi- the detail of neural architecture at present, it will tion of new reciprocally interacting subsystems. not be included in the following discussion on These developed as a result of both biological commensurability. A great deal more research and embodied cultural dynamics. New subsys- will be necessary to fill in the gaps in evidence, as tems depend on an iterative mechanism where well as to explore correspondences between our inputs coming from two sources (e.g.: audition methodology and the extant theories in CA. and body states feeding back changes in vocal In 1996, Mithen advanced an account for musculatures) are mapped together in multi- CA based on evolutionary psychology and the modal space, to establish their invariants that can modularity of mind argument (Barkow et al., bind them together (in this example, the invari- 1992; Buss, 2005; Pinker, 1997; Plotkin, 1997; ants that underlie vocal output and heard speech Fodor, 1983; Gardner, 1983). According to his is “phonology”). Once a repertoire of invariants model, human mind evolved from a series of differentiate from the bulk of other multimodal isolated domains of knowledge, also referred invariants, a new, functionally independent sub- to as “multiple intelligences”. These domains system emerges (Barnard et al., 2007). The pro- gradually became more interconnected with the posal is that an architecture of six interacting increasing complexity of the genus Homo, until subsystems can fully explain the behavioral rep- modern humans developed a module for meta- ertoire of apes, as well as that of our last common cognition. This had the ability to manipulate ancestor. Three additional modules, for vocal information flows between the other domains articulation, phonology and for propositional of intelligence. In this way, modern humans, but meaning, were added. This last addition brought not Neanderthals, evolved a fluid cognition that into existence not only propositional meaning, enabled them to innovate and develop the wide but also augmented precursor multimodal capa- range of Upper Paleolithic artifacts. bilities to effectively yield two levels of meaning. Coolidge & Wynn (2001, 2005, 2009) This in turn enabled a dialogue between the two advanced a proposal, based around a quantita- levels of meaning that support abstract thought tive aspect of Alan Baddeley´s tripartite working and innovation. Barnard´s mental architecture memory model. This architecture is composed can do more things at one and the same time of a central executive and two slave subsystems: than precursor architectures with fewer subsys- the phonological loop and the visuo-spatial tems. A nine-interacting subsystem architecture sketchpad (Baddeley & Hitch, 1974, Baddeley can walk, talk, chew gum and think at the same & Logie, 1999; Baddeley, 2000, 2001, 2003). time. Across the trajectory from six to nine sub- Coolidge & Wynn argued that selective advan- systems, the computational power of the full tage would have accrued to Homo sapiens with an architecture increased and with that the behav- increase in working memory capacity. This would ioral repertoires they were capable of exhibiting. D. GarofoliJ. Hawks & M. N. Haidle 27

Fig. 10 - Apparent isomorphism between the mental architectures, as drawn out from the a posteriori analysis of Mithen´s, Wynn & Coolidge´s and Barnard´s models.

There is good reason to argue that these men- comparable implemented theories. The effi- tal architectures are commensurable and could ciency of these resulting implemented theories be properly compared. Indeed, even though the can be judged in a two stage process. First, in nature of the Basic interactors [B]s varies among the construction of the core-theories, attention the various models, the architectures are con- would be focused on internal congruence of the structed by adopting the same logic. This sug- architectures, conformity to the experimental gested isomorphism between the three mental data, explanatory power during the horizontal architectures is shown in Figure 10. mapping, etc... Second, during the vertical map- In Mithen´s evolutionary psychology model, ping for the target-level, the resulting imple- the mental architecture [A] would be represented mented theories would be evaluated for their by a series of intelligences, which would act as plausibility. In this way, the number of ad hoc basic interactors [B]s and would be constrained hypotheses that need to be produced in order to eventually by a series of evolutionary determined support the explanations of the properties of the constituents [C]s. In Coolidge & Wynn´s pro- two missing systems (Fig. 5) will be counted. In posal, conversely, the basic interactors would be summary, these three commensurable theories covered by the subsystems in Baddeley´s model can be compared and analyzed through the cri- (visuospatial sketchpad, central executive, pho- teria for selection that we previously discussed. nological loop), composed themselves by sub- Thus, theories providing less efficient explana- components (for the phonological loop: articula- tions could be ultimately rejected. tory rehearsal and phonological store, perhaps), Unfortunately, apart from the small number while in Barnard´s model the mental architecture of implemented theories reported above, many would be represented by the interacting subsys- proposals in CA do not fit with our methodol- tems (the [B]s), which are composed themselves ogy. On the contrary, most of them appear to be by a number of constituent subcomponents (their largely incommensurable with those described internal structure), which stand for the [C]s. above. This general incommensurability has been If we consider, as we argued earlier, that an noted, for instance, by Wynn & Coolidge (2011), invariant behavioral architecture can be con- who highlighted the fact that ambiguous terms structed for these three different proposals, we like “complex” or “modern” cognition are widely have the possibility to map horizontally at least adopted nowadays by archaeologists interested two systems (i.e.: the mental and the behavio- in the evolution of mind and behavior. Indeed, ral) in a coherent way. As a consequence, we these notions are often used in place of precise have three different core-theories, which, albeit descriptions of the cognitive processes and men- only partial because of the absence of the neu- tal architectures necessary to produce artifacts (see ral architecture, can be used to produce properly also Dubreuil, 2011). For example, the elaborated

www.isita-org.com 28 Cognitive Archaeology

sequence of stages that are necessary for perform- exact nature of neural and cognitive architectures ing a Levallois reduction could be considered required to support the behavioral repertoire as a proof of underlying “complex” cognition. itself into consideration. For instance, defin- However, this tells us little, for instance, about ing the behavior that can be identified only in the working memory capacity that is needed to anatomically modern humans as modern tells perform the same tool-making task. As a con- us little about the “modernity” of the mental sequence, this difference in analytic categories architecture. Even if a particular set of behavioral clearly contributes to confusion within debates. practices is detected exclusively in Homo sapiens, Problems like these stem from the fact that in fact, the cognitive processes underlying them many theories in CA focus only on the proper- can be thoroughly consistent with a more primi- ties of the artifactual record. Thus, properties of tive mental architecture (Klein, 2003). a behavioral architecture, or even isolated frag- Problems like these become even more rel- ments of it, are used to directly draw inferences evant when we consider the relativistic drifts about an ill-defined structure of the mind (e.g.: that in turn plague the same notion of behavio- complex mind, modern mind, etc...). No micro- ral modernity. This is strictly dependent on the theories of the subcomponents of the mind itself authors´ beliefs and is not based on objective cri- and the macro-theory that rules their reciprocal teria (see Nowell, 2010 for an extensive review; interactions are taken into consideration. In our Soffer, 2009, p. 45). Modern behavior has been view, this leads to a series of situations like the largely associated with the use of symbols (Chase, first type of incommensurability, where logical 2003, 2006; Davidson & Noble 1989; Noble & connections and mappings are produced with a Davidson 1991; Henshilwood & Marean, 2003; theory-dependent logic. For instance, Barnard Gamble, 1999; Marean, 2007; Wadley, 2001), (2010b) reports that his diagram has a high risk but it could be possible to raise the threshold of being misused through what he defines as of the concept of modernity to agriculture, the “diagonal connections” between components of advent of writing systems, or even to commu- the architectures, as opposed to proper horizon- nication through the Internet (Chase, 2003). tal mapping. He discusses a case where specific Or perhaps we have never been modern, as properties of cognitive subsystems are inferred suggested by Latour in his homonymous book directly from isolated components of behavioral (1993). It appears then that it is extremely dif- architectures. As an example of improper diago- ficult to classify behavior under the qualitative nal connections, he uses the attempt to infer category of “modernity” and any assumption in properties of language from regularities in stone this direction inevitably holds a certain degree of knapping procedures (Holloway, 1969), which is arbitrariness, which grows exponentially when addressed without any horizontal reference to a transferred to the domain of cognition. theory of cognition. From these premises it follows that attempts Particular emphasis in discussing problems to infer the presence of a modern mental archi- of incommensurability should be given to the tecture from “modern” behavior are unsafe in spread use of the concept of “behavioral moder- principle. They ought to be replaced with proper nity” in the archaeological debate, an expression horizontal mapping, as proposed by Barnard´s that itself comes with a high degree of ambi- extended model. guity: there is no solid theory to account for it (Henshilwood & Marean, 2003). The main risk with attributing the label “modern” to a Case Studies behavioral architecture is that it can lead to the automatic transfer of this qualitative attribution Neanderthal symbolism to the mental and neural architectures as well. An example of incommensurable propos- Again, this can be performed without taking the als in the extant cognitive archaeological theory D. GarofoliJ. Hawks & M. N. Haidle 29

might be useful to build a clear perception of the perforated marine shells are uncontroversial in problems previously accounted for. Both Nowell UP and later prehistoric contexts but, as shown (2010) and dÉrrico & Stringer (2011) recently by the evidence from Africa, the Near East and reported on the most prominent schools of now Iberia, both behaviors first occur in the thought associated with the evolution of human MP/MSA. Their emergence in two continents, behavior and mind. Of the three current schools among two different lineages and, in the time of thought recognized by these authors, we find scale of human evolution, at about the same the situation with the “cultural” school (Chase, time, is inconsistent with cognitive-genetic 1999, 2003, 2006; Hovers et al., 2003, 2006; explanations and implies that these innovations dÉrrico et al., 1998; dÉrrico, 2003; Conard, were fulfilling a need—aiding in the personal 2008; Kuhn & Stirner, 2007; Zilhao, 2007) par- or social identification of people—that did ticularly interesting for our discussion on incom- not exist in the preceding two million years of mensurability. This school is known for claiming human evolution. Our findings therefore support that demographic changes in human populations models of the emergence of behavioral modernity might have been the main cause of variation in as caused by technological progress, demographic human behavior instead of cognitive or genetic increase, and social complexification and show factors. The rise in innovations is explained by that there is no biunivocal correlation between appealing to the growing number of inter-indi- “modern” anatomy and “modern” behavior (pp. vidual interactions within a wider community 13, 36–38).” (Shennan, 2001; Powell et al., 2009). A striking example that might be ascribed to this frame- From the quoted text, it seems clear that work is the recent Zilhao et al.´s (2010) work the authors are using beads to argue that on symbolic cognition in Iberian Neanderthals, Neanderthals posses the ambiguously defined associated with the findings of perforated beads trait of ´´behavioral modernity``, which in this in Mousterian layers. These artifacts resemble case is identified with symbolism on the base those found with modern humans in Africa of “growing consensus” (Marean, 2007, p. 367; and in the Near East during the Middle Stone see also Nowell, 2010). Behavioral modernity, Age (dÉrrico et al., 2005, 2008, 2009; d’Errico warranted by the use of beads, is thus used to & Vanhaeren, 2007; Bouzouggar et al., 2007; infer that - cognitive/genetic mechanisms are not Marean et al., 2007; Bar-Yosef et al., 2009.; necessary to produce those behaviors typical of Vanhaeren et al., 2006) as well as in the European fully modern humans only. On the contrary, the Upper Palaeolithic (Klein, 2008; Vanhaeren & authors conclude that even more archaic popu- d’Errico, 2006). lations can develop behavioral enhancements The authors discuss the implications of the by relying solely on demographic and social use of beads, which they consider to be sym- changes. However, no information is provided bols a priori, without providing a description of on which cognitive architecture is necessary or the semiotic relationship between objects, signs sufficient to produce body ornaments like beads. and interpretants (Peirce, 1839-1914 in Hoopes, Nor the authors clarify how this architecture 1991; Rossano, 2010; Deacon, 1997), which is influenced by the variation of demographic is desirable to precisely define the structure of and social dynamics in Neanderthal popula- behavior. Then, they draw directly conclusions tions. Furthermore, paleoneurological evidence about the cognitive level, as demonstrated in the (Bruner, 2003, 2004, 2008, 2010, 2011; Gunz, following passage taken from the same Zilhao et 2010), which would argue in favor of a cogni- al., 2010 (emphasis added): tive and biological hypothesis, is also not taken into consideration. This is even more evident “The symbolic implications of body painting and through the absence of any reference to a neural of the ornamental use of pigment-stained and architecture.

www.isita-org.com 30 Cognitive Archaeology

Fig. 11 - Mapping logic adopted by Zilhao et al. (2010). Modified with permission from Barnard (2010b), in Nowell & Davidson: Stone tool and the Evolution of Human Cognition. University Press of Colorado. The colour version of this figure is available at the JASs website.

In the light of our methodology, it can be differences were in play among modern humans argued that the logic adopted by Zilhao et al. and the non modern populations. (2010) is based on establishing connections The crucial point to grasp here is that the cul- between a part of a behavioral architecture, tural school´s conclusions could be contradict- an isolated B (use of beads in Neanderthals), ing the alternative proposals only by virtue of the directly to an improperly defined structure of the improper mapping adopted and not because they mind (A = symbolic). In Figure 11, indeed, we represent more plausible explanations. Once a can now visualize how a partial behavioral archi- proper mapping is adopted, the new constraints tecture is mapped on the whole structure of the offered by the many macro and micro-theoretical mind, without taking into account any macro- levels introduced can undermine the stability of theory or micro-theory for a mental architecture the same theory. In what follows, we will show (Incommensurability of the first type). how properly built core-theories could in prin- In addition, even if mapped with a mental ciple explain the use of beads by archaic popu- and a neural architecture, the logic is still incon- lations without supporting the cultural school gruent in constructing the architectures (incom- thesis of cognitive equivalence. mensurability of the second type) (Figure 12). Indeed, we can adopt Barnard´s extended As discussed before, single behavioral practices, diagram to construct a proper behavioral archi- considered in isolation from the entire behavioral tecture, which includes the use of beads as a basic repertoire, cannot be used alone to explain prop- interactor, along with all the repertoire of prac- erties of a cognitive architecture. tices and the constituent interactors that con- The inappropriate mapping adopted by Zilhao strain them. Then, if we map this onto a mental et al. (2010) leads to the claim that a qualitatively architecture and its subsystems, as previously modern cognitive architecture might have been described, and then on a neural architecture, we present in Neanderthals as well (Harrold et al., can realize that the cultural school conclusions 2009, p. 290) and discovered or exploited through does not follow necessarily from the use of beads. changing demographics/cultural phenomena. In fact, all the processes that are required to pro- This implemented theory radically contradicts the duce the Neanderthal behavioral architecture general conclusion shared by Mithen´s, Coolidge (poly-architecture), including those that underlie & Wynn´s and Barnard´s models, despite being the use of beads, could be supported by a non- incommensurable with them. These three pro- enhanced working memory, an architecture of posals, albeit with consistent theory-specific dif- eight interacting cognitive subsystems or a set of ferences, share the idea that cognitive/biological intelligences lacking metacognition (Coolidge & D. GarofoliJ. Hawks & M. N. Haidle 31

Fig. 12 - Example of an incoherent mapping produced by assembling incongruent architectures. Modified with permission from Barnard (2010b), in Nowell & Davidson: Stone tool and the Evolution of Human Cognition. University Press of Colorado.

Wynn, 2004; Wynn & Coolidge, 2004; Barnard, However, the idea that demographic changes can 2010a; Mithen, 1996). Contrary to the common lead to behavioral advancements, as hypothesized archaeological notion that symbolism is con- by Zilhao et al. (2010), can still be supported, if nected with beads, this behavioral practice could behavioral variations are performed within the be, in effect, reduced to a non-symbolic level limits of the capabilities of a non-fully-modern following a Saussarian framework (Gärdenfors, mental architecture. This, however, does not rule 2011), an enactivist perspective (Malafouris, out the fact that biological alterations might have 2007, 2008b), or more precisely to an indexical been necessary to reach a fully modern mental one, by adopting a Peircean semiotics approach architecture. In addition, it does not prove either (Rossano, 2010, 2011). In this way, cognitive that demographic changes alone are responsible processes at the base of indexical reference can for the “discovery” of cognitive capabilities in be identified, mapped on the mental architecture non-modern human species. and evaluated together with the rest of the behav- Nevertheless, we do not intend to say that ioral practices associated with Neanderthals. the approach of the cultural school cannot be By virtue of the universal laws that bind the followed or does not deserve any considera- architectures in the core-theory, the vertical map- tion a priori. Our goal was simply to compare ping then allows us to reconstruct the missing this school of thought with the other models architectures by presenting an explanatory set of described above, in order to discuss an example theories in the target level, which ends up in the of how theories can be incommensurable and formation of an implemented theory (Fig. 5). contradictory at the same time. Moreover, we In this case, if the use of beads does not exceed intended to point out the fact that some theories, those cognitive processes that can explain the rest once contextualized in a proper mapping frame- of the behavioral architecture too (see Coolidge work, can lose their original explanatory power. & Wynn, reply to Henshilwood & Dubreuil, Incidentally, it appears that all the models 2011), there is no need to argue in favor of a we have argued to be commensurable in our modern mental architecture in Neanderthals. methodology actually support the same final

www.isita-org.com 32 Cognitive Archaeology

conclusion. However, this does not imply that of this practice can be considered as broadly con- our methodology is necessarily linked to the sole sistent with our methodology. cognitive/biological explanatory models. This The authors provide a multidisciplinary might be an additional reason for the followers analysis of the controversial problem of the use of the “cultural” school to adjust their proposals of beads in Early-modern Homo sapiens popula- in a way that is commensurable with the other tions. They describe a behavioral architecture models. For example, they would need to specify that includes the production of beads, advance more clearly: a) what happens to mental and neu- explanations on the cognitive processes that ral architectures when the behavioral one varies are compatible with the use of beads as orna- in response to demographic and social changes, ments, and propose a neurological explanation b) which constraints prevent a modern cognitive for these cognitive requirements. Henshilwood architecture from being exploited in the absence & Dubreuil argue that beads can be considered of demographic changes, c) how current macro- as symbolic (p. 375) and that this explanation is theories of cognition explain the discovery of supported by a series of cognitive processes that latent cognitive processes, d) what the differ- implies the presence, within these populations, ence in parietal lobe anatomy between modern of a working memory capacity that almost equals humans and Neanderthals means (Bruner, 2004, that of contemporary humans (p. 379). This 2008, 2010) and so forth. On the other side, is furthermore supported by other synchronic the cultural school can contribute to our meth- archaeological evidence (pp. 372-375 and refer- odology by providing a framework to define an ences therein; Wadley, 2010). This hypothesis is assembled system of social and demographic also corroborated by their argument that there mechanisms/theories (Shennan, 2001; Powell et was a rise in complexity of the temporal lobes, al., 2009). This could exert further constraints which is in turn supported by studies in evolu- on the behavioral architecture and on the whole tionary neuroscience that show a disproportional process of generating core-theories. enlargement of the temporal volumes in modern Based on this approach, it would be possible humans as opposed to apes (Rilling & Seligman, to create candidate implemented theories, which 2002; Semendeferi & Damasio, 2000). In addi- might be properly compared with alternative tion to volumetric variations, they also refer to the theoretical proposals and evaluated for plausibil- enhanced functional characterization of temporal ity in light of our criteria for selection. In the cortices, their pattern of connectivity and devel- current state of affairs, the fact that the cultural opment (pp. 362-367 and references therein). school offers theories that are mostly incom- Comparing Henshilwood & Dubreuil´s pro- mensurable with those from the other schools posal with that of Coolidge & Wynn (2011), it of thought demonstrates that methodological can be argued that these authors refer, more or less relativism represents the status quo in cognitive directly, to the same set of theories in construct- archeology. This example embodies how deeply ing a core-theory (the same set of rules), which distorted (and inefficient) a theoretical debate might be associated to Baddeley´s framework based on contradictory and yet incommensura- (Baddeley & Hitch, 1974; Baddeley & Loogie, ble theories is. 1999; Baddeley, 2001, 2003). Even though this does not exactly match our proposal for produc- Use of beads in early modern humans ing assembled systems, their macro-theories and Recently, Henshilwood & Dubreuil (2011) the mapping processes, it is important to recog- advanced a proposal on a related topic, namely nize that we agree in the methodological direc- the use of beads in Early Modern human population that ought to be taken. It is worth noting tions in South-African Still Bay and Howieson’s that every assertion these authors make with Poort technocomplexes. Interestingly, their respect to behavior is connected to explanations approach to the potential symbolic implications in terms of executive functions and cognition, as D. GarofoliJ. Hawks & M. N. Haidle 33

well as neural substrates. This creates a series of be considered this way (Deacon, 1997). Without inferences that work as a holistic network. providing a detailed account on these points, the However, these groups of authors interpret the association between beads and symbolism seems behavioral architecture in different ways, so that to be unjustified. the relationships among the entities within the Henshilwood & Dubreuil defend their posi- behavioral architecture are different (one allows tion by stating that the symbolic value of beads symbolic exchange of information between enti- is explained by associating the abstract concept ties, the other only indexical). As a consequence, of “coolness” to these artifacts, as well as to con- the result of vertical mapping, along with the temporary ornaments like earrings, a notion that explanations drawn from the core-theory, both is nevertheless rejected by the former authors as differ. In this case, the same core-theory produces an ad hoc assumption. Henshilwood & Dubreuil different conclusions depending on how the might need a further ad hoc hypothesis to clarify behavioral architecture is constructed from the why a mental architecture, assumed to be capable artifactual record. The result of this vertical map- of handling metacognitive tasks in the creation of ping is the formulation of two candidate imple- symbols, kept this potential latent for thousands mented theories, which are perfectly congruent of years after reaching this level. From their side, in the rules of construction of the architectures, Wynn & Coolidge have to deal with problems internally coherent in the mapping processes traditionally connected with every mutation- and ultimately commensurable when compared. based proposal. For example, they need to pro- These two implemented theories, albeit contra- vide additional clarifications on the chronology dictory in their conclusions, can then be submit- of mutational events, the speed of replacement ted to our criteria of selection for plausibility. of alleles that specify for non-enhanced working Which conception of the behavioral archi- memory phenotypes, the outcome of behavio- tecture best fits with the artifactual record? How ral advantages, etc... (see for a review Wynn & many ad hoc hypotheses are requested to make Coolidge, 2011). the implemented theories stable? Do they leave some gaps of knowledge behind? Are they simple or is it necessary to multiply the assumptions Conclusions required to support them? Questions like these can be answered now by examining the differ- Our lack of access to the mind of ancient ence in the contents of the two implemented populations, as well as the necessity of recon- theories, not their rules of construction. structing its nature by mapping together mul- It is worth noting how the two groups of tiple and reciprocally interacting systems, repre- authors defend their respective models in the sent probably the most important epistemologi- context of the commentary to Henshilwood & cal issue in CA. Dubreuil´s work (2011). Coolidge & Wynn Scholars who have chosen to embrace a non- claim that Henshilwood & Dubreuil´s imple- reductionist view and to take these problems into mented theory is implausible, because ad hoc consideration might, therefore, be tempted to hypotheses are required to associate beads with overcome them by accepting methodological rel- symbolism and modern cognition. For instance, ativism. In this paper, however, we have argued to support the idea that beads are processed like that the intrinsic limits of this discipline do not true symbols, and not like indexes, it might be require scholars to adopt a subjective/interpreta- necessary to state clearly what the abstract con- tive approach. As an alternative to relativism, we cept embodied by these artifacts is, how this have proposed a method based upon a synthesis association is cognitively mediated, which cogni- of the following: tive processes are involved and how the symbol is 1) Horizontal Mapping - Meta-Theoretical inserted in a system of symbols that allows it to Level: The construction of universal laws of

www.isita-org.com 34 Cognitive Archaeology

connection between the interacting systems ex- would lead to the dangerous paradox of “formal- amined in CA (in our case neural, mental and izing” the idea that anything goes. To counter this behavioral architectures) based upon empiri- risk, we have proposed an extension of Barnard´s cal and theoretical research in extant species. (2010b) model. Our proposal provides a precise These rules should allow the prediction of logic to be followed when mapping domains of how any potential set of systems interacts, knowledge in CA. This method is still incomplete so that, given one system, and constraints and may include additional assembled systems. that govern their connections, the nature of Nevertheless, we argue that it is not only more the other two can be inferred. efficient than any relativistic perspective, but it is 2) Construction of the Target Level: The in- also necessary in order to allow candidate theories dividuation and definition of a behavioral to be properly compared. In this way, the risk of architecture, its own constitutive elements and comparing incommensurable theories, built from the way they interact. This operation is me- different epistemological bases, can be avoided. diated by the analysis of the archaeological Indeed, incommensurability can lead researchers record associated with the target species/ to use a subjective methodological perspective to population of interest. correct or ignore possible conceptual problems 3) Vertical Mapping - Implemented-Theory: that arise within a theoretical proposal. In addi- The deduction of missing architectures and tion, adopting a distorted methodology can lead their interactions in the target level. This is scholars to focus their attention and efforts on performed via universal principles of con- models that might be flawed in their contents. nection determined by point 1 above, as ap- These models can be perpetuated by virtue of this plied to the target level´s behavioral archi- methodological flaw, but can then divert effort tecture (point 2). As a result of this proce- and focus away from properly constructed models. dure, we have what we have defined an “im- Of course, it is not our position that all the plemented theory”. This is not necessarily studies in cognitive archaeology which are not valid in the explanations it provides, because aligned with our proposal should be totally dis- elements of implausibility can arise out of missed. Even works focusing on single or partial each stage of construction (point 1 and 2). architectures can, in fact, contribute valuable However, this resulting implemented theory new evidence, problems and questions concern- is logically coherent and can represent a ing the nature of the systems of interest. These candidate theory to be compared with other elements can be nonetheless contextualized candidate proposals in terms of plausibility. within our model and can contribute to the for- 4) Theory selection: Plausibility of candidate mation of more inclusive and more properly con- implemented theories is evaluated on the stituted theories. basis of criteria for selection and the most In addition, we are not claiming that our plausible theories selected given our state of approach is the only one admissible as a universal knowledge. methodology in this discipline. Instead, we hope to stimulate the development and the discussion This methodological proposal demonstrates of further techniques able to provide more effi- that, even in the absence of strict empirical refer- cient methodological tools than those discussed ence, it is still possible to adopt a methodological here. In our view two points are absolutely criti- framework that allows us to evaluate theoretical: i) providing efficient ways of establishing cal proposals on the base of rationally objective logical connections between/within assembled criteria, which markedly differ from subjective systems and ii) proposing rational criteria for interpretations. evaluating commensurable theories. These ele- In summary, within the wider field of CA ments represent the foundations of an anti-rela- remains a risk of reintroducing relativism, which tivistic agenda and must be adhered to. D. GarofoliJ. Hawks & M. N. Haidle 35

Proposals like ours are inevitably affected by mapping rules between systems). In the vertical the problem of dealing with real communities of mapping, this indicates both the propagation scientific practice. These are usually prone to self- of an incoherence in the horizontal mapping determination of their own methods and often phase, as well as the assumption of different unresponsive to potential suggestions of changes rules for mapping architectures between the from outside their communities (Bell, 1998). core and the target level. However, our short-term goals are two-fold. Incommensurability: This occurs when two or First, we hope that this paper will drive the atten- more implemented theories, each with differ- tion of cognitive archaeologists to the problems ent rules for mapping between and within sys- discussed here, encouraging new ideas. Second, tems, are built and thus cannot be compared we wish to provide a tool for identifying theories because of the differences in the methods of that carry elements of incommensurability and construction. that consequently need to be further developed. Incongruence: A term adopted to describe using of different rules of construction for single architectures, where the internal layers are built Glossary by referring to different levels of reference, so that each hierarchy is different than the others Architecture: An aggregated system of structural (i.e.: different in terms of the rules of construc- entities that interact by virtue of their proper- tion within each system). ties, creating a behavioral outcome. Level: The imaginary context where architectures Arcuate Fasciculus: A neural pathway that con- are aligned and mapped together by virtue of nects posterior regions of the brain (temporo- their properties. parietal junction) with the frontal cortex. Macro-theory: A theory that structures a system Core-theory: An abstract representation that of local theories in a holistic whole and not the defines a level where all the meta-theories are sum of reductive units. integrated together, in order to highlight the Meta-theory: A set of theories that are mapped invariant rules of system interaction that are together for a single species of interest and pre- valid for all the considered species. The result dicts how the constituent architectures interact is a theory that accounts for how neural, mental and constrain each other in every contingent and behavioral systems interact and constrain situation that characterizes that species. each other in all potential conditions, given the Micro-theory: A theory that is generally pro- evolutionary links that connect organisms. duced in experimental situations and accounts Horizontal mapping: The process of aligning for specific subcomponents of an aggregated architectures and establishing connections be- system/architecture. tween their constituent parts which follows a Post-processualism: A school of thought in ar- universal logic that is determined a priori. chaeological theory that denied the possibility Implemented Theory: The entire theoretical for archaeology to reach objective conclusions schema that results from the process of horizon- by adopting a scientific method. According to tal and vertical mapping and provides explana- the post-processualists, indeed, archaeology tions about the nature of ancient minds. was limited only to a subjective/interpretative Incoherence: In general, this term has been perspective. adopted to define a distorted mapping proce- Processual School: A movement in archaeologi- dure. More specifically, in the horizontal map- cal theory that supported the idea scientific ping, this is the result of mapping between in- method could be applied to archaeological re- congruent architectures or diagonal inferences search. Hypotheses could therefore be advanced that put layers that are not allowed to interact by collecting quantitative data and tested di- into communication (due to differences in the rectly from the archaeological record.

www.isita-org.com 36 Cognitive Archaeology

Vertical mapping: The process of connecting a Processing of Meaning. Curr. Anthropol., 51: behavioral architecture based on the artefactu- S39-53. al record to a core-theory, in order to produce Barnard P.J. 2010b. Current developments in in- explanations on the nature of extinct minds ferring cognitive capabilities from the archaeo- and brains. logical traces left by stone tools. In A. Nowell & I. Davidson (eds): Stone Tools and the Evolution of Human Cognition, pp. 207-226. University Acknowledgements Press of Colorado. Barkow J., Cosmides L. & Tooby J. 1992 (eds). We wish to thank Philip Barnard and David Duke The Adapted Mind: Evolutionary Psychology and for their helpful comments and suggestions on the the Generation of Culture. Oxford University theoretical foundations of the arguments exposed in Press, Oxford & New York. this paper. Any mistake remains our own. We thank Bar-Yosef Mayer D., Vandermeersch B. & Bar- also Philip Barnard, Iain Davidson, Krista Dotzel Yosef O. 2009. Shells and ochre in Middle and Alessio Garofoli for organizational and mate- Paleolithic Qafzeh Cave, Israel: indications for rial help during the preparation of this work. modern behavior. J. Hum. Evol., 56: 307–314. Bell J.A. 1987. Reason vs. relativism: review of I. Hodder’s Reading the past. Archaeological References Review from Cambridge. 6: 75-86. Bell J.A. 1991. Anarchy and archaeology. In R. Baddeley A.D. & Hitch G.J. 1974. Working W. Preucel (ed): Processual and postprocessual memory. In G.A. Bower (eds): Recent Advances archaeologies: multiple ways of knowing the past, in Learning and Motivation, pp. 47-90. pp. 71-80. Southern Illinois University Press, Academic Press, New York. Carbondale. Baddeley A.D. & Logie R.H. 1999. Working Bell J.A. 1994a. Interpretation and testability memory: the multiple-component model. In in theories about prehistoric thinking. In C. A. Miyake & P. Shah (eds): Models of Working Renfrew (ed): The Ancient Mind: elements of Memory: Mechanisms of Active Maintenance cognitive archaeology, pp. 15-21. Cambridge and Executive Control, pp. 28-61. Cambridge University Press, Cambridge. University Press, Cambridge. Bell J.A. 1994b. Reconstructing prehistory: scien- Baddeley A.D. 2000. The episodic buffer: a new tific method and archaeology. Temple University component of working memory? Trends Cog. Press, Philadelphia. Sci., 4: 417-423. Bell J.A. 1998. Overcoming dogma in epistemol- Baddeley A.D. 2001. Is working memory still ogy. Issues in integrative Studies, 16: 99-119. working? Am. Psychol., 11: 851-864. Binford L. 1965. Archaeological systematics and Baddeley A. 2003. Working memory: looking the study of culture process. Am. Antiquity., 31: back and looking forward. Nat. Rev. Neurosci., 203-210. 4: 829-839. Binford L. 1987. Data, relativism and archaeo- Barnard P., May J., Duke D. & Duce D. 2000. logical science. Man, 22: 391-404. Systems, interactions and Macro-theory. ACM Bouzouggar A., Barton N., Vanhaeren M., T. Comput. Hum. Int., 7: 222–262. d’Errico F., Collcutt S., Higham T., Hodge Barnard P., Duke D., Byrne R. & Davidson I. E., Parfitt S., Rhodes E., Schwenninger J.L., 2007. Differentiation in cognitive and emo- Stringer C., Turner E., Ward S., Moutmir A. & tional meanings: an evolutionary analysis. Stambouli A. 2007. 82,000-year-old shell beads Cognition Emotion, 21: 1155–1183. from North Africa and implications for the Barnard P.J. 2010a. From Executive Mechanisms origins of modern human behavior. Proc. Natl. Underlying Perception and Action to the Parallel Acad. Sci. U.S.A., 104: 9964-9969. D. GarofoliJ. Hawks & M. N. Haidle 37

Bruner E., Manzi G. & Arsuaga J.L. 2003. Conard N. 2008. A critical view of the evidence Encephalization and allometric trajectories in for a Southern African origin of behavioral mo- the genus Homo: Evidence from the Neandertal dernity. S. Afr. Arch. Soc., 10: 175-179. and modern lineages. Proc. Natl. Acad. Sci. Coolidge F.L. & Wynn T. 2001. Executive functions U.S.A., 100: 15335-15340. of the frontal lobes and the evolutionary ascendancy Bruner E. 2004. Geometric morphometrics and of Homo sapiens. Camb. Archaeol. J., 11: 255-60. paleoneurology: brain shape evolution in the Coolidge F.L. & Wynn T. 2004. A cognitive genus Homo. J. Hum. Evol., 47: 279-303. and neuropsychological perspective on the Bruner E. 2008. Comparing endocranial Chatelperronian. J. Anthropol. Res., 60: 55-70. form and shape differences in modern hu- Coolidge F.L. & Wynn T. 2005. Working memo- mans and Neandertal: a geometric approach. ry, its executive functions, and the emergence of PaleoAnthropology, 2008: 93-106. modern thinking. Camb. Archaeol. J., 15: 5–26. Bruner E. 2010. Morphological Differences in Coolidge F.L. & Wynn T. 2008. The role of epi- the Parietal Lobes within the Human Genus: a sodic memory and autonoetic thought in the Neurofunctional perspective. Curr. Anthropol., Upper Palaeolithic Life. PaleoAnthropology, 51: S77-78. 2008:212-217. Bruner E. 2011. The evolution of the parietal corti- Coolidge F.L. & Wynn T. 2009. The rise of Homo cal areas in the human genus: between structure sapiens. Wiley Blackwell, New Jersey. and cognition. In D. Broadfield, M. Yuan, K. Coolidge F.L. & Wynn T. 2011. Comment to C.S. Schick, & N. Toth (eds): The human brain evolv- Henshilwood & B. Dubreuil: The Still Bay and ing. Stone Age Institute, Bloomington, Indiana. Howiesons Poort, 77–59 ka Symbolic Material Bruner E., Martin-Loeches M., Burgualeta M. & Culture and the Evolution of the Mind during Colom R. 2011. Midsagittal brain shape cor- the African Middle Stone Age. Curr. Anthropol., relation with intelligence and cognitive perfor- 52: 361-400. mance. Intelligence, 39: 141-147. Davidson I. & Noble W. 1989. The archaeology Buss D.M. 2005. The Handbook of Evolutionary of perception: traces of depiction and language. Psychology. Wiley Press, Hoboken, New Jersey. Curr. Anthropol., 30: 125–156. Chalmers D.J. 1995. Facing Up to the Problem of Davidson I. & Noble W. 1993. Tools and Language Consciousness. Journal of Consciousness Studies, in Human Evolution. In K. Gibson & T. Ingold 2: 200-19. (eds): Tools, Language and Cognition, pp: 363- Chalmers D.J. 1996. The Conscious Mind: in search of a 388. Cambridge University Press, New York. fundamental theory. Oxford University Press, Oxford. Davidson I. 2010. The archaeology of cognitive Chase P.G. 1999. Symbolism as reference and evolution. WIREs Cog. Sci., 1: 214-229. symbolism as culture. In R.I.M. Dunbar, C. Deacon T. 1997. The symbolic species. Norton & Knight & C. Power (eds): The Evolution of Company Inc., New York. Culture: An Interdisciplinary View, pp. 34–99. Deregowski J.B. 2002. Is symmetry of stone Edinburgh University Press, Edinburgh. tools merely an epiphenomenon of similarity? Chase P.G. 2003. Comment on C.S. Henshilwood Comment to T. Wynn.: Archaeology and cogni- & C. Marean: The origin of modern behavior: a tive evolution. Behav. Brain Sci., 25: 389–438. review and critique of models and test implica- d’Errico F., Zilhao J., Julien M., Baffier D. & tions. Curr. Anthropol., 44: 637. Pelegrin J. 1998. Neanderthal acculturation in Chase P.G. 2006. The Emergence of Culture: The Western Europe? A critical review of the evi- Evolution of a Uniquely Human Way of Life. dence and its interpretation. Curr. Anthropol., Springer, New York. 39: 1-44. Clark A. 2008. Supersizing the mind: Embodiment, d’Errico F. 2003. The invisible frontier. A multiple action and cognitive extension. Oxford University species model for the origin of behavioral mo- Press, Oxford. dernity. Evol. Anthropol., 12: 188-202.

www.isita-org.com 38 Cognitive Archaeology

d’Errico F., Henshilwood C., Vanhaeren M., van Gunz P., Neubauer S., Maureille B. & Hublin J.J. Niekerke K. 2005. Nassarius kraussianus shell 2010. Brain development after birth differs be- beads from Blombos Cave: evidence for sym- tween Neanderthals and modern humans. Curr. bolic behaviour in the Middle Stone Age. J. Biol., 20: R921-R922. Hum. Evol., 48: 3-24. Haidle M.N. 2009. How to think a simple spear? d’Errico F. & Vanhaeren M. 2007. Evolution or In S. de Beaune, F.L. Coolidge & T. Wynn revolution? New evidence for the origins of (eds): Cognitive archaeology and human evolu- symbolic behavior in and out of Africa. In P. tion, pp. 57-73. Cambridge University Press, Mellars, K. Boyle, O. Bar-Yosef & C. Stringer New York. (eds): Rethinking the Human Revolution., Haidle M.N. 2010. Working memory capacity pp. 275–286. McDonald Institute for and the evolution of modern cognitive capaci- Archaeological Research, Cambridge. ties - implications from animal and early hu- d’Errico F., Vanhaeren M. & Wadley L. 2008. man tool use. Curr. Anthropol., 51: S149-S166. Possible shell beads from the Middle Stone Age Haidle M.N. 2012. How to think tools? A com- layers of Sibudu Cave, South Africa. J. Archaeol. parison of cognitive aspects in tool behavior of Sci., 35: 2675–2685. animals and during human evolution. Cognitive d’ Errico F., Vanhaeren M., Barton N., Bouzouggar perspectives in tool behaviour Vol. 1. http:// A. & Mienis H. et al. 2009. Additional evi- tobias-lib.uni-tuebingen.de/frontdoor. dence on the use of personal ornaments in the php?source_opus=6014. Middle Paleolithic of North Africa. Proc. Natl. Halford G.S., Cowan N. & Andrews G. 2007. Acad. Sci. U.S.A., 106: 16051–16056. Separating cognitive capacity from knowledge: d’Errico F. & Stringer C.B. 2011. Evolution, revo- a new hypothesis. Trends Cog. Sci., 11: 236-242. lution or saltation scenario for the emergence of Harrold F. 2009. Historical perspectives on the modern cultures? Philos. Trans. R. Soc. Lond. B., European transition from Middle to Upper 366: 1060-9. Paleolithic. In M. Camps & P. Chauhan (eds): Dubreuil B. 2011. The other middle-range theo- Sourcebook of Paleolithic Transitions, pp. 283– ries: Mapping behaviour and the evolution of 299. Springer, New York. the mind. In C.S. Henshilwood & F. d’Errico Henshilwood C.S. & Marean C. 2003. The origin (eds): Homo Symbolicus: The dawn of language, of modern human behavior. Curr. Anthropol., imagination and spirituality, xi, 237, pp. 185– 44: 627–651. 204. John Benjamins Publishing Company, Henshilwood C.S. & Dubreuil B. 2011. The Still Amsterdam, The Netherlands. Bay and Howiesons Poort, 77–59 ka Symbolic Fodor J.A. 1983. The Modularity of Mind: An Material Culture and the Evolution of the Essay in Faculty Psychology. The MIT Press, Mind during the African Middle Stone Age. Cambridge, Massachusetts. Curr. Anthropol., 52: 361-400. Gamble C. 1999. Paleolithic Societies of Europe. Hodder I. 1984. History vs science: no contest. Cambridge Univ. Press, Cambridge. Review of L.R. Binford (1983): In pursuit of Gardner H. 1983. Frames of Mind: the theories of the past (Thames and Hudson) and J.G.D multiple intelligences. Basic books, New York. Clark (1983): The identity of man (Methuen). Gärdenfors P. 2011. Comment to C.S. Henshilwood Scot. Archaeol. Rev., 3: 66-68. & B. Dubreuil: The Still Bay and Howiesons Hodder I. 1986. Reading the past. Cambridge Poort, 77–59 ka Symbolic Material Culture and University Press, Cambridge. the Evolution of the Mind during the African Hodder, I. 1987. Reply to Bell, J.A.: Reason vs. rela- Middle Stone Age. Curr. Anthropol., 52: 361-400. tivism: review of Ian Hodder’s “Reading the past”. Godfrey-Smith P. 2003. Theory and Reality: Archaeological Review from Cambridge, 6: 75-86. An Introduction to the Philosophy of Science. Holloway R.L. 1969. Culture: a human domain. University of Chicago Press, Chicago, Illinois. Curr. Anthropol., 10: 395-412. D. GarofoliJ. Hawks & M. N. Haidle 39

Hoopes J. 1991. Peirce on Signs. Writing on semi- Latour B. 1993. We Have Never Been Modern. Harvard otics by Charles Sanders Peirce. The University University Press, Cambridge, Massachusetts. of North Carolina Press, Chapel Hills, North Lieberman D.E., McBratney B.M & Krovitz G. Carolina. 2002. The evolution and development of cra- Hovers E., Ilani S., Bar-Yosef O. & Vandermeersch nial form in Homo sapiens. Proc. Natl. Acad. B. 2003. An Early Case of Color Symbolism. Sci. U.S.A., 99: 1134-1139. Curr. Anthropol., 44: 491-522. Lombard M. & Haidle M.N. 2012. Thinking Hovers E. & Belfer-Cohen A. 2006. Now you see a bow-and-arrow: cognitive implications of it, now you don’t—modern human behavior Middle Stone Age bow and stone-tipped arrow in the Middle Paleolithic. In E. Hovers & S.L. technology. Camb. Archaeol. J., 22: 237-264. Kuhn (eds): Transitions Before The Transition: Malafouris L. 2007. Before and beyond represen- Evolution and Stability in the Middle Paleolithic tation: towards an enactive conception of the and Middle Stone Age, pp 205–304. Springer, Palaeolithic image. In C. Renfrew & I. Morley New York. (eds): Image and Imagination: a Global History Hume D. [1739] 1978. A Treatise of Human of Figurative Representation, pp. 289–302. Nature. In L.A. Selby-Bigge & P.H. Nidditch. McDonald Institute Monographs, Cambridge. Oxford University Press, Oxford. Malafouris L. 2008a. Between brains, bodies Johnson M. 1999. Archaeological Theory: An and things: tectonoetic awareness and the ex- Introduction. Blackwell, Oxford. tended self. Philos. Trans R. Soc. Lond. B, 363: Jung R.E. & Haier R.J. 2007. The parieto-frontal 1993–2002. integration theory (P-FIT) of intelligence: con- Malafouris L. 2008b. Beads for a Plastic Mind: verging neuroimaging evidence. Behav. Brain The ‘Blind Man’s Stick’ (BMS) hypothesis and Sci., 30: 135–187. the active nature of material culture. Camb. Klein R.G. 1995. Anatomy, behavior, and modern Archaeol. J., 18: 401-414. human origins. J. World Prehist., 9: 167–198. Malafouris L. 2008c. At the Potter wheel. In C. Klein R.G. 2000. Archaeology and the evolution Knappet & L. Malafouris (eds): Material agency, of human behavior. Evol. Anthropol., 9: 17–36. pp. 19-36. Springer editions, New York. Klein R.G. 2001. Southern Africa and modern Malafouris L. 2010a. The brain–artefact interface human origins. J. Anthropol. Res., 57: 1–16. (BAI): a challenge for archaeology and cul- Klein R.G. 2003. Comment to E. Hovers, S. Ilani, tural neuroscience. Soc. Cogn. Affect. Neurosci., O. Bar-Yosef & B. Vandermeersch: An Early doi: 10.1093/scan/nsp057 Case of Color Symbolism. Curr. Anthropol., 44: Malafouris L. 2010b. Metaplasticity and the hu- 491-522. man becoming. Principles of Neuroarchaeology. Klein R.G. 2008. Out of Africa and the evolu- J. Anthropol. Sci., 88: 49-72. tion of human behavior. Evol. Anthropol., 17: Marean C.W. 2007. Heading north: an Africanist 267–281. perspective on the replacement of Neanderthals by Kline M.A. & Boyd R. 2010. Population size modern humans. In P. Mellars, K. Boyle, O. Bar- predicts technological complexity in Oceania. Yosef & C. Stringer (eds): Rethinking the Human Philos. Trans R. Soc. Lond. B, 277: 2559-2564. Revolution. pp. 367–79. McDonald Institute for Kohn M. & Mithen S. 1999. Handaxes: products Archaeological Research, Cambridge. of sexual selection? Antiquity, 73: 518–526. Marean C.W., Bar-Matthews M., Bernatchez J., Kuhn S. & Stiner M. 2007. Paleolithic orna- Fisher E., Goldberg P., Herries A.I.R., Jacobs ments: implications for cognition, demography Z., Jerardino A., Karkanas P., Minichillo T., and identity. Diogenes, 54: 40–48. Nilssen P.J., Thompson E., Watts I. & Williams Lakoff G. & Johnson M. 1999. Philosophy in the H.M. 2007. Early human use of marine re- flesh: the embodied mind and its challenge to west- sources and pigment in South Africa during the ern thought. Basic Books, New York. Middle Pleistocene. Nature, 449: 905–908.

www.isita-org.com 40 Cognitive Archaeology

Mitchell J.C. 1995. Studying biface butchery fishing groups. Retrieved from: http://www.es- at Boxgrove: roe deer butchery with replica cholarship.org/uc/item/61n4303q handaxes. Lithics, 16: 64–69. Renfrew C. 1982. Towards an archaeology of Mithen S.J. 1996. The prehistory of the mind : a mind, pp. 16-19. Cambridge University Press, search for the origins of art, religion, and science. Cambridge. Thames and Hudson, London. Renfrew C. 1989. Comments on archaeology into Mithen S.J. & Parsons L. 2008. The brain as a cul- the 1990s. Nor. Archaeol. Rev., 22: 33-41. tural artifact. Camb. Archaeol. J., 18: 415-422. Renfrew C. 1994. Towards a cognitive archaeol- Neubauer S., Gunz P. & Hublin J.J. 2010. ogy. In C. Renfrew (eds): The Ancient Mind: Endocranial shape changes during growth in elements of cognitive archaeology, pp 3-12. chimpanzees and humans: a morphometric Cambridge University Press, Cambridge. analysis of unique and shared aspects. J. Hum. Rilling J.K. & Seligman. R. 2002. A quantita- Evol., 59: 555–56. tive morphometric comparative analysis of Newell, A. 1990. Unified theories of Cognition. the primate temporal lobe. J. Hum. Evol., 42: Harvard University Press, Cambridge, 505–534. Massachusetts. Rossano M. 2010. Making Friends, making Tools and Noble, W. & Davidson, I. 1991. The evolution- making Symbols. Curr. Anthropol., 51: S89-98. ary emergence of modern human behavior: Rossano M. 2011. Comment to Henshilwood, language and its archaeology. Man, 26: 223–53. C.S. & Dubreuil, B.: The Still Bay and Nowell A. 2010. Defining Behavioral Modernity Howiesons Poort, 77–59 ka Symbolic Material in the Context of Neandertal and Anatomically Culture and the Evolution of the Mind during Modern Human Populations. Ann. Rev. the African Middle Stone Age. Curr. Anthropol., Anthropol., 39: 437–452. 52: 361-400. Nowell A. & Davidson I. 2010. Stone Tools and the Samson R. 2006. Stones of contention: the evolution of human cognition. University Press of Acheleuan handaxe lethal projectile controver- Colorado. sy. Lithic Technol., 31: 127-135. Peikoff L. 1981. May be you are wrong. The Searle J. 1997. The Mistery of Consciousness. New Objectivist Forum, 2: 8-12. York review books, New York. Pinker S. 1997. How the Mind Works. Norton, Searle J. 2004. Mind: a brief introduction. Oxford New York. University Press, New York. Platt J.R. 1964. Strong Inference. Science, 146: Semendeferi K. & Damasio H. 2000. The brain 347–353. and its main anatomical subdivisions in living Plotkin H. 1997. Evolution in Mind. Alan Lane, hominoids using magnetic resonance imaging. London. J. Hum. Evol., 38: 317– 332. Powell A., Shennan S. & Thomas M.G. 2009. Shanks M. & Tilley C. 1987a. Re-constructing archae- Late Pleistocene demography and the appear- ology. Cambridge University Press, Cambridge. ance of modern human behavior. Science, 324: Shanks M. & Tilley C. 1987b. Social theory and ar- 1298–1301 chaeology. Cambridge Polity Press, Cambridge. Preucel R. 2007. Archaeological Semiotics. Shennan S. 2001. Demography and cultural in- Blackwell publishing, Malden, Massachusetts. novation: a model and its implications for the Rand A. 1964. The virtue of selfishness: a new con- emergence of modern human culture. Cambr. cept of egoism, pp. 13-35. Signet, New York. Archaeol. J., 11: 5–16. Rand A. 1967. Introduction to objectivist epistemol- Soffer O. 2009. Defining modernity, estab- ogy. The Objectivist, New York. lishing rubicons, imagining the other—and Read D. 2012. Population size does not pre- the Neanderthal enigma. In M. Camps & dict artifact complexity: analysis of data from P. Chauhan (eds): Sourcebook of Paleolithic Tasmania, Arctic hunter-gatherers, and Oceania Transitions, pp. 43–64. Springer, New York. D. GarofoliJ. Hawks & M. N. Haidle 41

Trigger B.G. 2007. A History of Archaeological Wynn T. & Coolidge F. 2004. The expert Thought (Second Edition). Cambridge Neandertal mind. J. Hum. Evol., 46, 467–487. University Press, New York. Wynn T., Coolidge F. & Bright M. 2009. Vanhaeren M., d’Errico F., Stringer C., James Hohlenstein-Stadel and the Evolution of S.L., Todd J.A. & Mienis H.K. 2006. Middle Human Conceptual Thought. Camb. Archaeol. Paleolithic shell beads in Israel and Algeria. J., 19: 73–83. Science, 12: 1785–1788. Wynn T. & Coolidge F.L. 2011. The Implications Vanhaeren M. & d’Errico F. 2006. Aurignacian of the Working Memory Model for the ethno-linguistic geography of Europe revealed Evolution of Modern Cognition International. by personal ornaments. J. Archaeol. Sci., 33: J. Evolution. Biol, Article ID. 741357. 1105-1128. Zilhao J. 2007. The emergence of ornaments and Wadley L. 2001. What is cultural modernity? A art: an archaeological perspective on the origins of general view and a South African perspective behavioral modernity. J. Archaeol. Res., 15: 1–54. from Rose Cottage Cave. Cambr. Archaeol. J., Zilhão J., Angelucci D.E., Badal-García E., d’Errico 11: 201–21. F., Daniel F., Dayet L., Douka K., Higham T.F.G., Wadley L. 2010. Were snares and traps used in the Martínez-Sánchez M.J., Montes-Bernárdez R., Middle Stone Age and does it matter? A review Murcia- Mascarós S., Pérez-Sirvent C., Roldán- and a case study from Sibudu, South Africa. J. García C., Vanhaeren M., Villaverde V., Wood R. Hum. Evol., 58: 179–192. & Zapata J. 2010. Symbolic use of marine shells Walton D.N. 2001. Abductive, presumptive and mineral pigments by Iberian Neandertals. and plausible arguments. Informal Logic., 21: Proc. Natl. Acad. Sci. U.S.A., 107: 1023-1028. 141-169. Wynn, T. 2002. Archaeology and cognitive evolution. Behav. Brain Sci., 25: 389–438. Associate Editor, Emiliano Bruner