Polyagentive archaeology works from the understanding of two realities of the world; the virtual and the actual.
The virtual
(V1) The non-empirical and immanent level of polyagentive archaeology is the level of virtual (qualitative) multiplicities (Bergson 1998, 2001, 2004; Deleuze 1991, 1994). The virtual is a nonnumeric duration which is impossible to split up. It is pure quality. The virtual is the past that is simultaneous with the present, which it actualizes through becomings. It is always turning into something else by differentiating and repeating. This virtual multiplicity becomes actual (quantitative) multiplicities that make up the analytical and spatialized world. It is the latter we find as polyagents, which are actualized objects with “causative capabilities” in an actualized polyagentive network.
(V2) Polyagency is a collection of intensive processes that lie in-between the virtual and the actual. It is a distributed agency that lacks an identity of its own, but it generates other identities through the becoming, both externally and internally (Grosz 2001). It is rhizomatic, meaning that there is no genealogy here, no straight line in the polyagentive process. It is in the in-between where individuation takes place, where virtuality closes onto itself and forms an actualized boundary to what is external.
(V3) Both polyagency and virtuality lack metric spatiality (space seen in Euclidean terms) and actualized/spatialized temporality (when time is seen as sharing the characteristics of space). However, they generate polyagents that have spatiality (Grosz 2001). With polyagentive archaeology it is possible to use both “long-term” and instantaneous perspectives of the same data. The virtual that persists in a “long-term” perspective is actualized through events that come down to us as material patterns.
(V4) The virtual is the ontological foundation for a complex ontology of ontologies (Aijmer 2001; Wittgenstein 1998). No ontology can be said to be truer than any other. However, the virtual can be found in all ontologies, but it does not explain how their actualizations are perceived by human beings which can be reached by other ontological perspectives. This “virtual ontology of actual ontologies” focuses on temporal movement rather than on substance and representation.

The actual
(A1) The empirical level of the concept is the level of actual multiplicities; in another word – the polyagents that are divided into materialities and intangibilities. From a physical sense, these consist of a union of singularities. A singularity is an intensity where something specific is concentrated and differentiated. Materiality consists of singularities that changes human behaviour, such as when liquid water turns into ice. A frozen lake can be used differently than a non-frozen lake. This is what separates materiality from “material culture”, materiality is not just artefacts or buildings. Polyagents are what we can account for, but they are static representations of the becoming. Originally, polyagency only meant “causative capabilities” of polyagents in Gell’s sense of agency (Normark 2004). This notion is still maintained, but only in the actualized level. The word agency in the concept excludes intentions, will or mind. It mainly relates to that which is active and not static.
(A2) Polyagents share similarities with Latour’s (2000) actants. However, the focus is on the material polyagents, but intangible ones can sometimes be studied. These interact with each other and the way these actualizations intersect with each other is also polyagency because the encounter generates becomings. Since the encounter also diverges tendencies within the virtual, this rhizomatic interaction between actualized objects forms a discontinuous “genealogy” that has to be recreated and constructed from a human perspective (Bachelard 2000; Foucault 1979, 1984).
(A3) The polyagentive (“machinic”) assemblage consists of several polyagents that together form a unity. The polyagentive (“machinic”) phylum is a polyagentive lineage that is a constellation of singularities (Deleuze and Guattari 1988). Any polyagent that is part of a polyagentive lineage can develop into other phyla. The phyla and assemblages cut through each other through the virtual processes and therefore they do not represent static entities. DeLanda (2002) places the machinic phylum in the virtual since it is formed through virtual processes. I place the polyagentive assemblage and phylum in the actual since they can only be defined by materialities.
(A4) The actual is the level of representations (language, signs and materiality/territory). Here a semiotics with no meaning is used. The sign/representation/materiality is only seen as an index or prototype of polyagents. This indexical polyagent only needs the human agent to become reproduced (Gell 1998). The idea of the distributed indexical polyagent is used to explain how a polyagentive phylum is reproduced. A collection of various distributed indexical polyagents forms a polyagentive oeuvre that is unique for every locale or site. An oeuvre is the total amount of indexical polyagents at a site or locale at a specific time.
(A5) The indexical polyagents, phyla, assemblages and oeuvres are nested together in polyagentive networks of various spatio-temporal extents. The network also relates to non-indexical polyagents. It has nodes that connect other networks at other spatial locations. The network is temporal and is constantly fluctuating and brings along virtualities from near and far to various nodes of the network. It is up to the researcher to define the nodes of interest. Each node is a polyagent within a phylum, such as causeways of the Maya area.
The return of the human becoming
Although the intentions of past human agents are of secondary concern to this approach, the human agents are introduced at a later stage in the analysis, as reproducers of virtual ideologies and actual ideologies. The virtual ideology is directed towards matter that affects actual ideologies. Actual ideologies focus on forms and style and therefore create representations such as symbols, architecture and writing. The actual ideology ranges from individual habits and nomadic thought to arbolic macro-ideologies. The tendencies of the virtual ideology are always working, changing and diverging whereas the actual ideologies it intersects with are perceived as static and ontologically secure. This is a way to describe how what appears to be continuity from the Middle Formative of the Maya area to present day (such as the quadripartite principle in “Maya cosmology”) mainly is continuity in the virtual tendencies of materiality, and not in the transcendental actual ideologies Mayanists usually deal with (Normark 2006a, 2006b, 2007, in preparation).
The human agents are still part of the archaeological investigation, but they cannot be separated from the “non-human”. They are no longer the centre of attention and action (Pickering 1995:6). The polyagent cannot be fully understood if we do not include manufacturer, user and non-user of the polyagent (Latour 1993). We need a polyagentive network. This polyagentive network is not completely symmetrical. The symmetry between “material agency” and human agency breaks down when it comes to intentionality. Pickering argues that human beings and machines (and in my view other material objects as well) are intertwined in a tuning process, but that there is no strict symmetry. Human beings have a planning capacity, set up goals and make plans whereas materiality does not. These plans emerge temporally but they transform in the encounters with materiality (Pickering 1995:17-18). Therefore, Pickering disagrees with Latour’s idea that there is symmetry between human beings and non-humans. For Latour there is no primacy between different forms of agents before the encounter (Bruun Jensen 2003b:230). Pickering rather suggests a dualism that is dependent on human intentionality. Human action often has future goals in view (Bruun Jensen 2003a:87). In the polyagentive approach, it is argued that the past human beings indeed would have been of primary importance if we could study them at first hand, but since they are lacking, they need to be decentralized.
How is a polyagent defined in its actual state? Since it derives from the actualization of the virtual, it means that each soil particle in the ground is a polyagent. Although this is true, that is not an operational archaeological unit, unless we are studying soil particles. Polyagency shares with the virtual the lack of a defined temporal and spatial location. Polyagency is a space without a space. It is a relationship that lies in-between. If an artefact is broken, its polyagency is not broken. The polyagency has just internally differentiated and relates to two objects or actualizations that each have different tendencies, but that maintain the same original virtuality. It is up to the researcher to define the spatial and temporal extent of a polyagent which is either a material or an intangible object. This has to be defined in each case.
My dissertation primarily concerns one form of polyagents; causeways (sakbeob) found in the Maya area and particularly those at the two neighbouring Mexican sites of Ichmul and Yo’okop. The actual states of their causeways always fluctuate in their contours. The virtual tendencies of the causeways keep them united from the points of view of particular actual ideologies; archaeologists see them as archaeological structures and road constructors see them as potential construction fill for new roads. These are examples of how materialities affected and still affect people’s actions. However, the ways we view them are as actualized multiplicities of a virtuality. As such they can have a multitude of humanocentric explanations (actual ideologies) which are developed in contact with a virtual ideology that is immanent to materiality.
Although, the polyagentive archaeology sees the objects as active, it does not become a fetishism since the object is not seen to carry with them anything concerning ethnicity, gender, practice or cosmology like humanocentric archaeologists believe. We cannot know discourses (actual ideologies) of the past from materiality itself, but we could possibly reach the immanent iconic codes as Aijmer (2001) suggests. The iconic codes are similar to my virtual ideology that focuses on the immanent rather than the transcendent. However, that is beyond the scope of this text.
References
Aijmer, G.
2001 The symbological project. Cultural Dynamics 13(1):66-91.
Bachelard, G.
2000 The Dialectic of Duration. Translated by M. McAllester Jones. Clinamen Press, Manchester.
Bergson, H.
1998 Creative Evolution. Dover Publications, Mineola, New York.
2001 Time and Free Will. Dover Publications, Mineola, New York.
2004 Matter and Memory. Dover Publications, Mineola, New York.
Bruun Jensen, C.
2003a Interview with Andrew Pickering. In Chasing Technoscience: Matrix for Materiality, edited by D. Ihde and E. Selinger, pp. 83-95. Indiana University Press, Bloomington & Indianapolis.
2003b Latour and Pickering: post-human perspectives on science, becoming, and normativity. In Chasing Technoscience: Matrix for Materiality, edited by D. Ihde and E. Selinger, pp. 225-240. Indiana University Press, Bloomington & Indianapolis.
DeLanda, M.
2002 Intensive Science and Virtual Philosophy. Continuum, London.
Deleuze, G.
1991 Bergsonism. Zone Books, New York.
1994 Difference and Repetition. Columbia University Press, New York.
Deleuze, G. and F. Guattari
1988 A Thousand Plateaus: Capitalism and Schizophrenia. Athlone, London.
Foucault, M.
1979 Discipline and Punish: The Birth of the Prison. Penguin, Harmondsworth.
1984 Nietzsche, genealogy, history. In The Foucault Reader, edited by P. Rabinow, pp. 32-50. Peregrine, Harmondsworth.
Gell, A.
1998 Art and Agency: An Anthropological Theory. Clarendon, Oxford.
Grosz, E.
2001 Architecture from the Outside: Essays on Virtual and Real Space. MIT Press, cop, Cambridge, Mass.
Latour, B.
1993 We Have Never Been Modern. Harvester Wheatsheaf, New York.
2000 When things strike back: a possible contribution of ‘science studies’ to the social sciences. British Journal of Sociology 51(1):107-123.
Normark, J.
2004 Caught Somewhere in Time: Polyagentive Archaeology in the Maya Lowlands. GOTARC Series C, Volume 52. Department of Archaeology, Göteborg University, Göteborg.
2006a Ethnicity and the shared quasi-objects: Issues of becoming relating to two open-fronted structures at Nohcacab, Quintana Roo, Mexico. In Maya Ethnicity: The Construction of Ethnic Identity from the Preclassic to Modern Times, edited by F. Sachse, pp. 61-81. Acta Mesoamericana, Volume 19. Verlag Anton Saurwein, Markt Scwaben.
2006b The Roads In-Between: Causeways and Polyagentive Networks at Ichmul and Yo’okop, Cochuah Region, Mexico. GOTARC Series B, Volume 45. Department of Archaeology, Göteborg University, Göteborg.
2007 Lethal encounters: Warfare and virtual ideologies in the Maya area. In Encounters/Materialities/Confrontations: Archaeologies of Social Space and Interaction, edited by P. Cornell and F. Fahlander. Cambridge Scholars Publishing, Cambridge.
in preparation The triadic causeways of Ichmul: virtual highways becoming actual roads. In 2500 Years of Occupation in the Cochuah Region (tentative title), edited by J. M. Shaw. University of Arizona Press, Tucson.
Pickering, A.
1995 The Mangle of Practice: Time, Agency and Science. University of Chicago Press, Chicago.
Wittgenstein, L.
1998 Filosofiska undersökningar. Natur och Kultur, Stockholm.