Ritual Theory and Attitudes to Agency in Brazilian Spirit Possession1
Steven Engler
Department of Humanities, Mount Royal College, 4825 Mount Royal Gate SW,
Calgary, AB T3E 6P8
E-mail: sengler@mtroyal.ca
Abstract
This article works with theory of ritual in order to begin addressing a series of questions
raised by Brazilian spirit possession rituals (in Kardecism and Umbanda). Four
contributions to theory of ritual highlight relevant conceptual issues: Humphrey and
Laidlaw on non-intentionality; Bloch on deference; Houseman and Severi on social
relations; and Kapferer on virtuality. Strawson’s philosophical distinction between
objective and reactive attitudes toward intentionality is used to make a case (i) that
certain formal aspects of ritual (indexicals) serve to (ii) mark culturally-variable
attitudes to agency within rituals, which are related to, but fundamentally distinct from,
non-ritual attitudes to agency.
Keywords
Ritual, theory of ritual, agency, indexicality, Umbanda, Kardecism, social relations,
Brazil
Recent theory of ritual has been moving past linguistic communication models that
emphasize the symbolism of rituals and that attempt to read their syntax and semantics.
This article makes a specific theoretical contribution that draws together themes from
recent work on theorizing rituals. The first section of the article offers a brief overview
of two Brazilian religions, Umbanda and Kardecism. Highlighting the interrelations
between spirit possession and social relations raises a series of questions. The
theoretical perspective set out in the following sections suggests a means of answering
these questions. The second section sketches four contributions to theory of ritual:
Humphrey and Laidlaw on non-intentionality; Bloch on deference; Houseman and
Severi on social relations; and Kapferer on virtuality. The third section draws on the
work of philosopher Peter Strawson to distinguish two distinct attitudes toward agency,
reactive and objective. This article’s most significant contribution is to argue that rituals
help to reframe the relation between these attitudes in a manner distinct from their
relation in non-ritual contexts. By focusing on this relational tension we turn our
attention from reading the meaning of ritual to the structure and dynamics of ritual
practice and how these reframe agency. I illustrate the ramifications of this distinct by
examining its relations with the four contributions to theory of ritual discussed in the
previous section. The fourth section draws critically on the work of Alfred Gell to argue
that the Peircean concept of indexicality is invaluable in analyzing which elements and
aspects of ritualized action achieve these attitudinal effects. The resulting theoretical
frame emphasizes agency, virtuality and indexicality in ritual within specific social
contexts. This will be illustrated, in the fifth and final section, by an analysis of spirit
possession in Kardecism and Umbanda.
	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
   	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  
1

Research for this article was funded in part by Brazil’s Coordenação de Aperfeiçoamento de Pessoal de
Nível Superior (CAPES). I am indebted to Jens Kreinath for suggesting the value of focusing on
indexicality, for introducing me to the work of Alfred Gell, and for many engaging and insightful
conversations. Translations from Portuguese are mine.

1	
  

I make three general claims. First, the balance between reactive and objective
attitudes to agency is transformed in the case of ritualized actions. Second, this
constitutes in important aspect of ritual virtuality, of i.e., of ritual’s ability to create an
altered set of social relations, “a space in which participants can reimagine (and redirect
or reorient themselves) into the everyday circumstances of life” (Kapferer 2004: 47).
Third, the fact that certain elements of ritual function as indexes (in the Peircean sense)
is crucial to this process. That is, some aspects of ritualized actions can only be
understood within, and insofar as they point to, their specific context. The words ‘this’
and ‘now’ are indexical because their meaning, and the truth-value of sentences in
which they are found, can only be fully determined with sufficient knowledge of the
context in which they are uttered. My thesis is as follows: certain sorts of ‘thisness’
found in spirit possession rituals in Brazil alter the tension between reactive and
objective attitudes; this naturalizes the agency of supernatural agents; and the virtuality
of these rituals both rehearses and inserts this transformed agency into broader social
relations of reciprocity.
Spirit Possession and Social Relations in Kardecism and Umbanda
Spirit possession is important in several areas of the Brazilian religious landscape: folk
religion; Pentecostalism and Neo-Pentecostalism; the Catholic charismatic revival;
Afro-Brazilian religions; Kardecism or Spiritism, a French-import with roots in
nineteenth-century American Spiritualism; and Umbanda, a twentieth-century mixture
of Afro-Brazilian religions and Kardecism. I focus on the latter two, though related
claims could be argued regarding the others. Kardecism and Umbanda manifest a range
of possession phenomena, including spirit mediumship of different types, voluntary
cultic possessions that involve long-term relationships between a given medium and a
given spirit, involuntary possession by these same or similar spirits of ritual participants
and observers, and malevolent possession, which is treated as an important cause of
physical and mental illness. I focus on spirit mediumship and voluntary cultic
possessions, which manifest a clearer relation to the broader field of Brazilian social
relations. A brief overview of these two religions will raise questions that highlight the
potential value of certain conceptual and theoretical resources.
Kardecism is a descendent of nineteenth-century French Spiritualism (Bastide
1967; Aubrée and Laplantine 1990; Hess 1991; Sharp 2006; Monroe 2008) Umbanda is
a mixture of Kardecism with Afro-Brazilian traditions (Brumana and Martinez 1989;
Brown 1994). Kardecism is an urban, middle-class religion, with a highly educated
membership among Brazilian religions (Pierucci and Prandi 2000: 633). Umbanda is a
distinctively Brazilian religion, formed in the 1920s and 1930s as a mixture of AfroBrazilian traditions and Kardecism. Although census results are especially problematic
in Brazil, given the high degree of multiple adherence and syncretism, the basic
numbers give us a useful sense of the scope of these two religions (Jacob et al. 2003:
101-105). In the 2000 census, 2.2 million Brazilians self-identified as Kardecists and
397,000 as Umbandists. (Candomblé, the largest of the Afro-Brazilian traditions, is
much smaller, with only 118,000 Brazilians claiming this as their primary religious
affiliation.)
Kardecism’s beliefs include reincarnation, karma, the universal spiritual
evolution of humankind, the practicability and value of communication with more
evolved spirits (including Jesus Christ), a plurality and hierarchy of worlds, a
transcendent God, and as an exceptionally evolved spirit. Key rituals include textual
study sessions, consultation with or reception of messages from spirits received by
2	
  

mediums (through oral communication, automatic writing or other means), the “passe”
(a form of blessing similar to New Age cleansing of the aura), and rituals of
“disobsession” conducted to free people from the pernicious influences of confused or
un-evolved spirits.
Three distinctly Brazilian characteristics of Kardecism stand out. First, since its
inception, Kardecism in Brazil has been more religious than in France, where its
“empirical” and scientific nature was emphasized (Warren 1968: 397; Machado 1983:
114; Stoll 1999: 41). Second, it has reflected Brazilian racial beliefs, above all rejecting
black and native spirits as non-evolved, a key element in the tensions that led to the
emergence of Umbanda as a separate religion in the early twentieth-century (Brown
1985: 11). Third, it posits a specifically Brazilian conception of the superior, more
highly evolved nature of the spirit guides and the worlds or “colonies” in which they
live (Xavier 2006a; 2006b; see F.L. Silva 2005). In its new home, Kardecism received a
distinctly Brazilian emphasis, reflecting specific concepts of religion, race, and social
hierarchy.
The origin of Umbanda is closely tied to issues of race and class. This fact is
basic to understanding how it came to be that Umbanda occupies a spectrum of
doctrinal and ritual positions between Kardecism and the Afro-Brazilian traditions like
Candomblé, with individual groups varying widely in their balance between the
characteristics of these two extremes (Engler forthcoming). Three tendencies, reflecting
the Brazilian myth of three races (DaMatta 1987: 58-85), were present in the formation
of Umbanda and continue to influence its development (Brown 1977; Brown and Bick
1987).
First, Spiritists looked to Afro-Brazilian traditions for a more intensely
emotional and corporeally satisfying symbolism and ritual, leading to the
empretecimento of Kardecism (Ortiz 1999, 40-45). This view is reflected in historical
accounts that focus on the role of white middle-class kardecists as leaders of the
merging religion, portraying Umbanda as originating in a middle-class empretecimento
of Kardecism.
Second, analyses that focus on social and demographic factors place greater
emphasis on the movement in the other direction, from Afro-Brazilian traditions
towards Kardecism (Brown 1977). Candomblé was the most important Afro-Brazilian
tradition in this context. Its key rituals include initiation, divination, and the roda-desanto (saint wheel) in which initiated members dance counter-clockwise, to intensely
syncopated drumming, until they enter into a trance state, becoming cavalos (horses) for
the orixás. This model of possession remains prominent in some Umbanda terreiros but
is not found in Kardecism. Two factors led, in the late nineteenth century, to the
embranquecimento of Candomblé and other Afro-Brazilian traditions, primarily in Rio
de Janeiro and São Paulo: (i) the influence and Spiritism and elements of Catholicism
on the lower classes, which the led to the formation of a “low Spiritism”; and (ii) the
presence of increasing numbers of white members, often new immigrants, in the AfroBrazilian traditions (Camargo 1961, 34-35; Ortiz 1999, 34-40). Edison Carneiro’s study
of Afro-Brazilian religions in Bahia, primarily in the 1930s, for example, found spiritist
ideas being absorbed into Candomblé de Caboclo, a tradition already incorporating both
African and indigenous elements (Carneiro 1977: 73-74; Ortiz 1999, 36).
Third, indigenous figures played an important role in the emergence of
Umbanda. Kardecist mediums began to receive the spirits of Brazilian Indians in the
1920s. These caboclos was rejected by some as less evolved. Their presence continues
to be an important marker distinguishing Kardecism from Umbanda (Concone 2001;

3	
  

Motta de Oliveira 2007). Caboclos are also present in some Afro-Brazilian traditions
(Boyer 1992; Santos 1992, 1995; Harding 2005: 122; Prandi 2005: 121-138).
There are a number of significant differences between Kardecism and Umbanda
(Camargo 1961; Aubrée and Laplantine 1990: 179-184; Brown 1994: 15-25; V.G. Silva
2005: 99-127; Lewgoy 2006). In general, Kardecism is predominantly a middle-class
religion. It is much more centralized institutionally and formally coherent in doctrine
and practice. It has a strong emphasis on textuality, with highly literate members, a
larger range of publications, and a central role for spirit writing as a ritual form. AfroBrazilian religions tend to be lower class, widely diffuse in belief and practice, and
orally based. Umbanda occupies a spectrum between Kardecism and the Afro-Brazilian
traditions in all these senses. With few institutionally imposed or maintained norms,
individual Umbanda terreiros manifests a spectrum of doctrinal and ritual
characteristics, from Umbanda branca (“white” Umbanda), that resembles Kardecism,
to a more Africanized Umbanda that resembles Candomblé and other Afro-Brazilian
religions. The former has a higher proportion of white and middle-class members and
the latter includes a greater proportion of black and lower-class members. (It is
important to keep in mind that race in Brazil is a complex issue, with tensions less
sharply defined than in other areas of Latin America [Lovell and Wood 1998; Hoffman
and Centeno 2003; Fischer 2004])
On the other hand, there are important ritual similarities between the two
religions, including the predominance of women as participants (though of men in
leadership roles), the passe or laying on of hands (more common in Kardecism), and the
blessing of items such as T-shirts and medallions, given to infants and the sick for the
purposes of protection and cure (more common in Umbanda).
There are several important differences, in general, between rituals in Kardecism
and Umbanda. In most of these cases, Umbanda terreiros at the “white” or Kardecist
end of the spectrum manifest these differences to a lesser extent and those at the African
end of the spectrum to a greater extent.
First, during possession rituals, Kardecist mediums remain fully or partially
conscious of themselves as distinct from the spirits that they receive, whereas, in
Umbanda, the medium, or cavalo, normally enters a classic trance state, sometimes so
deep as to require additional rituals to end the possession.
Second, Kardecist mediums receive “evolved” spirits, often middle-class
professionals like doctors, lawyers and intellectuals; they reject black and native spirits
as non-evolved. In Umbanda, the two most common types of spirits are marginal
figures: caboclos and preto-velhos (indigenous and black spirits respectively, central in
this religion but ‘marginal’ in terms of dominant social evaluations and access to socioeconomic opportunities). Participants are also often possessed by orixás (the divinities
of Candomblé) and by exus, a type of trickster or demonic figure present in Candomblé
but which seldom possesses members of that religion.
Third, Kardecist rituals are often text-centered, unlike Umbanda. Doctrinal
knowledge is central. Most Centers have a bookstall that sells Kardecist literature,
including the books currently being used in the weekly study sessions.
Fourth, where Kardecists will often return home from the Center with a reading
assignment, Umbandist are often told by the spirits they consult to perform additional
rituals at home, often involving lighting candles in a carefully prepared environment
with symbolic substances, colors and the presence of running water.
Fifth, Kardecists have a more fluid and democratic hierarchy. Meetings are
generally led by a senior male, echoing Protestant congregations. Newcomers begin by
attending study sessions and only slowly graduate to attending rituals where spirits are
4	
  

received or patients disobsessed. Umbandists have a more rigid hierarchy of
mediumship and a sharp line, doctrinally and as inscribed in the ritual space, between
clients and mediums.
Sixth, Kardecists dress like they were going to a Christian church. Umbandist
clients prefer white and avoid black. Mediums are distinguished by more elaborate
forms of dress, with variations that mark different types and levels of mediums. These
markers are more obvious at the African end of the spectrum where mediums place
greater emphasis on donning the clothing and taking up the artifacts associated with the
particular type of spirit that possesses them.
Seventh, Umbanda more explicitly reflects Brazilian social relations. Peter Fry
argues that Umbanda reflects the social and political structures of Brazilian society:
Umbanda is plausible insofar as the personal relations established with the
spirits, in hopes of obtaining favours, are homologous with the real relations
established for people’s benefit in the broader social system. ... Umbanda ... is a
ritual dramatization of the principles that govern life in the large cities of
Brazil.... Umbanda is a ritualized and dramatized metaphor that refers to Brazil’s
social and political reality (1978: 45, 47).

Concone notes that the religion’s various spirits “are obviously drawn from the national
reality” (2001: 282). Brumana and Martinez characterize Umbanda as a “subaltern cult”
that “elaborates symbolically the social condition of the client” (1989: 45). Ortiz argues
that “umbandist ideology preserves and transforms Afro-Brazilian cultural elements
within a modern society, [while, at the same time] manifesting rupture, forgetting, and
reinterpretation of older, traditional values” (1999[1978]: 212). Patricia Birman
underlines Umbanda’s symbolic and ritual engagement with Brazilian social reality:
“umbandist possession … is worthy of credit to the extent that it contextually invokes
its relation with the world as experienced by its audience” (1995: 44-45). This complex
refraction of Brazilian social relations in Umbanda is not found to the same extent in
Kardecism, where issues of hierarchy work themselves out more in the traditional
rhetoric of institutional politics. That is, in Kardecism’s hierarchical and meritocratic
view of spiritual progress is reflected in the relative absence of tensions between
personal and impersonal modes of authority.
Eighth, and finally, this tendency of Umbanda to more closely reflect tensions
within the hierarchical relations of Brazilian society, while at the same time standing
apart from the economic sphere, is correlated with its greater openness to a range of
conceptions of gender and sexuality (Landes 1947; Fry 1982; Birman 1982; 1985; 1995;
Natividade and Oliveira 2007).
Several issue emerge as targets for potential explanation and interpretation. The
historical importance of issues of race and class in relations between Kardecism and
Umbanda, and the parallels between aspects of these religions and specific
characteristics of Brazilian social relations underline the need to foreground these
contextual dimensions. Differences between the two religions, in terms of class
position, leadership and institutional structures, literacy, textuality, types of possession
states, etc. might make sense in terms of some relation to this context; on the other
hand, differences in ritual form must be kept clearly in mind.

5	
  

Marshalling Theoretical Resources

The view that ritual is communication has important precedents, e.g., Durkheim, Leach,
van Gennep, Geertz, and Turner. This broad set of approaches holds that ritual conveys
meaning like language does: by referring—albeit less literally, more symbolically—to
something other than itself. This approach has trouble dealing with the efficacy and
dynamics of rituals as well as with discrepancies among insider interpretations of rituals
and between those of insiders and scholars. As a result, key recent works emphasize
formal aspects of ritual (e.g., Laidlaw and Humphrey 2006; Kreinath 2006; Severi
2006). Fritz Staal marked an extreme: “A widespread but erroneous assumption about
ritual is that it consists in symbolic activities which refer to something else. It is
characteristic of a ritual performance, however, that it is self-contained and selfabsorbed. … Ritual is pure activity, without meaning or goal” (Staal 1979: 3, 9).
As I tried to make sense of Brazilian spirit-possession rituals, I found many
traditional approaches to theorizing rituals to be problematic. Issues like who is agent
and who patient, who sender and who receiver, who performer and who audience
opened up fascinating dimensions of ambivalence. My attempts to make sense of what I
saw were challenged by the dynamism of the rituals and the variety of interpretive
positions that reflect and shape them. After much reading and thought, I found a number
of recent theoretical approaches to be especially valuable: certain features of the case,
seemed best characterized in terms of certain key concepts, and these in turn seemed
more amenable to certain theoretical approaches. My approach to working with theory
of ritual converged on what Jens Kreinath, Jan Snoek, and Michael Stausberg call
“Theorizing Rituals,” a theoretically pluralistic approach, drawing on multiple
perspectives in a methodologically open and reflexive manner, and that foregrounds a
dynamic interplay between paradigmatic concepts and theoretical affiliations (2006; cf.
Engler 2008: 25-27). Four approaches emerged as particaulr useful.
First, Caroline Humphrey and James Laidlaw make the relation between ritual
and intentionality central to “contrasting ritualized action with action which is not
ritualized” (1994: 2). Less radical than Staal, they hold that rituals are learned purely as
sequences of actions and that meanings are applied to them after the fact. They argue
that ritual transforms the relation between action and intentionality: the “ritualization of
action” … consists in it becoming non-intentional, stipulated, and elemental or
archetypal” (Laidlaw and Humphrey 2006: 278) That is, rituals effectively disengage
the form of intentionality associated with action in non-ritual contexts: ritual actions are
not intended in the same way that non-ritual actions are. This reflects the fact that rituals
stipulate a sequence of actions that stands prior to the individual ritual actor. This results
in its perceived elemental or archetypal quality. The ‘non-intentionality’ of ritual is the
first concept I draw on below.
Second, Maurice Bloch helped to clarify this link between non-intentionality and
the archetypal quality of ritualized action. He agrees that intentionality is displaced
when action is ritualized: “any act … that appears to originate fully from the actor
cannot properly be called ritual” (Bloch 2006: 496). He clarifies this by noting that
rituals involve extensive quotation or deference, that is, “reliance on the authority of
others to guarantee the value of what is said or done” (Bloch 2006: 497). In this light,
tradition serves as a “phantasmogoric quasi-person” that guides ritual regardless of
individual intentions (Bloch 2006: 504). Bloch echoes Humphrey and Laidlaw
6	
  

regarding the non-intentional, stipulated, and archetypal quality of ritualized action, but
he is more specific regarding the cognitive mechanism involved. He also offers a more
explicit bridge to discussions of one of the psychological functions of ritual: “when one
is in trouble and does not know what to do, one allows oneself to be taken over by the
knowledge and the authority of others” (Bloch 2006: 506). This provides a second
insight that I will draw on: displacement of intentionality is not simply characteristic of
ritual in a descriptive sense; it has functional and even strategic dimensions.
Third, like Humphrey and Laidlaw, Michael Houseman and Carlo Severi (1998)
focus on the organizational features of ritual action, not on issues of meaning. More
specifically, they focus on the social context that ritual creates: “[I]t is the form of the
relational field in which the protagonists are engaged which underlies the establishment
of a context specific to ritual behaviour” (Houseman and Severi 1998: 167). They argue
that ritual acts out special relationships, and that this is what sets it apart from nonritualized action: “To the degree that ritual performances incorporate … exceptional
situations, they become readily recognizable as distinct from everyday interaction: they
cannot be fully accounted for in terms of ordinary internalities and patterns of
relationship” (Houseman 2006: 418-419). This third insight places the first two other in
a social context, raising the possibility that shifts in intentionality both reflect and
reconstitute social relations.
Fourth, Bruce Kapferer goes further in specifying the relationship between ritual
and non-ritual fields of social relations. He similarly focuses on “ritual as a technical
practice rather than a representational formation” (2006: 672). He argues that ritual, due
to its ‘virtuality,’ allows participants to exercise their agency in
a self-contained imaginal space—… a construction that enables participants to
break free from the constraints or determinations of everyday life…. The
phantasmagoric space of ritual virtuality may be conceived as a space … whose
dynamic not only interrupts prior determining processes but also ... [within]
which participants can reimagine (and redirect or reorient themselves) in the
everyday circumstances of life (Kapferer 2006: 673-674; cf. 2004: 47).
Ritual is not subordinate to non-ritual nor does it simply represent it. Ritual’s difference
allows participants to act differently; it similarity allows them to bring that distinct
agency into play in the non-ritual sphere. This fourth insight draws together the others
in suggesting that the distinct social field constituted in and through ritual can be
characterized by its virtuality, not entirely unlike yet not reducible to the sphere of nonritual relations, and providing a space within which participants reimagine, redirect or
reorient themselves. In analyzing Brazilian spirit-possession rituals, I will characterize
this re-orientation in terms of alternative conceptions of agency and intentionality.
In sum, these four recent theoretical approaches to ritual jointly emphasize
several valuable themes: ritualization (that which makes ritual distinct from non-ritual
action); a correlated emphasis on the form and dynamics of ritual (as opposed to its
meaning); a distinct relationship between intentionality and action (as compared to nonritual actions); a quality of authority that ritualized action gains in part due to this
transformation of intentionality; a distinct field of social relations constituted within and
by rituals; and the virtuality of the ritual sphere, autonomous and providing a space
where participants reimagine themselves, especially their agency and intentionality.
	
  

	
  

Reactive and Objective Attitudes to Agency

7	
  

Theorizing possession rituals in terms of the four theoretical insights above gives
us a general starting point: ritual creates a social space, like and unlike that of non-ritual
action, in which distinct forms of agency, characterized by the displacement of agency,
play themselves out. The next step is to clarify how ritual agency is distinct. Houseman
and Severi (1998: 231-232) critique Humphrey and Laidlaw’s idea that ritual action is
non-intentional as being negative and residual. What does ‘non-intentional’ mean? It is
not enough to simply assert that ritualized action is characterized by prior stipulation:
that might make some sense of highly formalized and scripted rituals, like many in
South India, but a more general approach to ritual needs to say something more concrete
about the intentional/non-intentional distinction. There are types and degrees in the
tempering or displacement of intentionality. Describing cases of how intentionality is
transformed can clarify the relation between agency and ritual dynamics, presenting a
spectrum of stances toward intentionality that ritual participants occupy and draw upon.
My main contribution in this article is to offer a conceptual distinction that
shows promise in clarifying the senses of agency involved. It is an initial step, drawing
on a specific set of cases, but with potential value more broadly. Brazilian spiritpossession rituals offer a case where something quite specific seems to be going on, in
terms of the types of agency of the various ritual participants. In this section, then, I
propose a distinction between two attitudes toward the actions of oneself and others: the
first recognizes a normal relation between intentionality and action; and the second
responds to an abnormal quality of this relation.
Philosopher Peter Strawson, in a classic essay, “Freedom and Resentment”
(1976), argued for a middle ground on the thorny issue of free will vs. determinism by
distinguishing two distinct attitudes toward agency. Strawson notes that our attitudes
towards the actions of others, when these actions affect us directly, vary along a
spectrum between two extremes. We adopt a ‘reactive’ attitude when we judge the
actions of the person who has benefited or harmed us to have been intentional, willed,
self-directed, etc. We adopt an ‘objective’ attitude to the extent that we judge that
person to have acted unknowingly, unintentionally, or in some other manner outside
their conscious control. In the former case, we experience the full range of appropriate
emotional reactions (Strawson mentions gratitude but focuses on resentment). In the
latter case, we acknowledge the person’s causal role, but we temper our reactions based
on our judgment that the usual relation between intentionality and action has been
somehow transformed.
On this view, our attitudes toward the agency of others (and toward our own
agency in hindsight) depend upon our evaluation of relations between action and
intentionality, between agency and causality. In some cases, we recognize that people
are not fully the agents of their actions. Strawson focuses on cases of the ‘not guilty by
reason of insanity’ sort:
the participant attitude, and the personal reactive attitudes in general, tend to
give place, and it is judged by the civilized should give place, to objective
attitudes, just in so far as the agent is seen as excluded from ordinary adult
human relationships by deep-rooted, psychological abnormality (Strawson 1976;
cf. Yearly 1985).
Interestingly, he notes, in passing, examples comparable to possession rituals:
we may think of such statements as ‘He wasn’t himself’, ‘He has been under
very great strain recently’, ‘He was acting under post-hypnotic suggestion’….
8	
  

Such pleas as these do … invite us to suspend our ordinary reactive attitudes
towards the agent, either at the time of his action or all the time. … They invite
us to view the agent himself in a different light from the light in which we
should normally view one who has acted as he has acted. I shall not linger over
… [such] cases (Strawson 1976).
I propose to do what Strawson did not, to linger over such cases, exploring the
theoretical leverage that his distinction offers in the case of spirit possession.
In general terms, religion is often closely associated with tensions between free
and constrained action. More specifically, we can note that religion offers both cases
where it is appropriate to adopt an objective attitude due to the non-intentionality of an
agent, e.g., due to altered states of consciousness, and also cases where it is appropriate
to adopt an objective attitude due normative constraints, e.g., where an agent is required
to act according to moral or ritual codes.
The distinction between reactive and objective attitudes is relevant to a broader
range of cases than those that Strawson discusses. Because his focus is on issues of free
will, responsibility, punishment, and moral condemnation, he examines only cases
where the disconnect between intentionality and action, between agency and causality,
is internal and non-stipulated. That is, he focuses on cases where the agent is free from
morally relevant constraint except insofar as their intentionality becomes unhooked
from their actions due to ignorance, immaturity, psychological problems, or altered
states of consciousness. In this type of case, voluntary and willed actions are the norm,
and involuntary and unwilled actions are abnormal. Hence, a reactive attitude is the
norm and an objective attitude the exception.
As a result, regarding this class of actions, Strawson’s distinction functions in
two ways. First, it frames certain cases of human actions as abnormal, tempering or
sustaining our normal responses and attitudes. This has clear implications in spheres
like law, psychiatry and childrearing. In the case of religion, there are obvious cases
where normal agentic link between intention and action is transformed in comparable
ways. For example, in spirit possession it is appropriate to adopt an objective attitude
toward the agency of the person who is possessed (though perhaps a reactive attitude to
the spirit or other agent that is possessing the patient). However, the specific attitude—
from reactive to objective—that is seen as appropriate in such rituals will vary
depending on a number of factors, including the type and degree of possession and a
wide variety religious and social beliefs and practices. For example, the predominance
of women as participants in ritual possession invokes complex and widely varying
questions of agency and power that bear on judgments of praiseworthiness, blame, and
the associated issues of status (see, e.g., Boddy 1994: 415-422; Rasmussen 1994;
Caciola 2000; Chesnut 2003; Hayes, 2006).
The second function of the distinction between reactive and objective attitudes is
that the latter normalizes the former. That is, the existence of cases that invoke an
objective attitude, of cases where we judge that people are not responsible for their
actions, helps to delimit, define, normalize and naturalize cases of “normal” agency, i.e.,
where we adopt a reactive attitude. As Steven Yearley notes, in exploring the
implications of Strawson’s distinction for legal and scientific reasoning, “in the law all
explanations are concerned with making sense of departures from willing compliance
with legal rules…. [C]orrect … [action] is treated as self-explanatory and thus in no
need of special investigation” (Yearley 1985, 117-118).
Strawson’s analysis meets his purposes. However, it can be enlarged in two
ways for theory of ritual. First, we can take into account that agents form reactive and
9	
  

objective attitudes toward their own actions in retrospect. For example, it is appropriate
to adopt an objective attitude toward our own action, as well as those of others, in cases
of altered states of consciousness, in cases of moral or ritual obligation, or where
individual agency is seen as radically subsumed under divine agency. Phrases like “In
ša’ Allāh” or “God willing, for example,” mark beliefs that subtly displace reactive
attitudes, sacralizing mundane contexts though this transformation: by delimiting our
own agency, we emphasize God’s.
Second, we can take account of another important and related category of cases,
on that Strawson does not address given his interest. We adopt objective rather reactive
attitudes in the face of others’ actions where those actions are previously stipulated or
normatively constrained, for example, in cases of obeying the law, respecting moral
rules or customs, keeping promises, carrying out bureaucratic processes, and, of course,
performing rituals. (Recalling Humphrey and Laidlaw’s emphasis on prior stipulation
and the resulting non-intentionality of ritual action, the relevance of Strawson’s
distinction is obvious.) Our reactive attitudes are tempered when people are perceived
as doing what they must or ought to do, as opposed to acting purely on the basis of their
own unfettered intentionality. That is, we adopt objective attitudes to a greater or lesser
extent when people act under various forms of normative constraint. That is, we find it
harder to resent or praise someone who is simply following orders or just doing their
job.
Such cases are contrary to those that Strawson considers in an important sense.
In the cases of non-intentionality that he considered, reactive attitudes were normal and
objective attitudes exceptional. However, where actions are previously stipulated or
normatively constrained, the opposite is true: acting according to these constraints, even
against one’s will, is the norm, and acting freely and without constraint is less usual. A
more objective attitude is the norm in the sort of cultural and institutional contexts that
frame the bulk of our actions in society, and a more reactive attitude is the exception. To
paraphrase Yearley, acting as normatively stipulated is self-explanatory and thus in no
need of special investigation. Both the types of case discussed by Strawson and this
second type associate the objective attitude with limitations on agency. In the case of
non-intentional actions, emphasized by Strawson, reactive attitudes are suspended due
to the involuntarily faulty agency of the actor. In the case of normatively constrained
actions, as is often typical of religion, reactive attitudes are tempered due to the
voluntary subordination of the actor’s independent agency to those norms. Strawson
considers only the former case, but both are relevant to theory of ritual.
In effect, Strawson proposes a linear model (see Figure 1a), in which the
spectrum from reactive to objective attitudes is correlated with a distinction between
freely chosen and fully intended actions, on the one hand, and actions characterized by
non-intentionality on the other, i.e., a break between intention and action due to
ignorance, psychological abnormality, etc. Taking into account previously stipulated or
normatively constrained actions adds a second independent variable (see Figure 1b). As
either the non-intentionality of an action or the normative constraints on it increases, so
will our tendency to adopt an objective attitude toward it.

10	
  

Reactive

Objective

Nonintentionality

Figure 1a. Objective attitudes to agency as a function of nonintentionality.

Objective

Reactive

Normative
Constraints	
  

Figure 1b. Objective attitudes to agency as a function of normative
constraints.
This distinction between attitudes toward agency has implications for our
assessment of Humphrey and Laidlaw’s claim that ritualized action is characterized by
non-intentionality. They are right to draw attention to the non-intentional character of
ritual action, but this concept is too broad. By itself, non-intentionality fails to
distinguish ritual action from the kinds of cases that Strawson draws our attention to
(non-intentionality by virtue of faulty agency). Their additional claim that ritual action
is stipulated and archetypal effectively sets it apart from those cases. However, as just
noted, ritualized action is only one of various sorts of action characterized by normative
constraints.
Bloch’s concept of deference adds an important dimension. According to Bloch,
ritual has three distinguishing characteristics. The first two are deference (i.e., “reliance
on the authority of others to guarantee the value of what is said or done” [2006: 497])
and consciousness of deference. In the case of spirit possession, for example, “the act of
deferral takes center stage, and everybody joins with … the medium in abandoning their
intentionality and in making themselves transparent to whomever’s words they are
quoting, which strangely fade out of focus” (Bloch 2006: 501). This far, Bloch covers
much the same ground as Humphrey and Laidlaw’s criteria that ritual action is nonintentional, stipulated and archetypal. His third criterion takes us further: “lack of clarity
on the person to whom one is deferring” or “indetermination of the originating mind”
(2006: 500, 502):
The secret to the problem of wanting to locate meaning without having normal
originators of that meaning is to merge all the shadowy transparent figures into
11	
  

one phantasmagoric quasi-person who may be called something like ‘tradition’,
‘the ancestors as a group’, ‘our way of doing things’, ‘our ‘spirit’, our ‘religion’,
perhaps even ‘God’ (Bloch 2006: 502, 504).
On Bloch’s account, “deference is a common aspect of human life,” but ritual is
distinctive because it “involves high degrees of deference”: “Rituals are orgies of
conscious deference” (Bloch 2006: 505-506). The emphasis on the consciousness of
deference and the link between non-intentionality and the legitimation of authority are
valuable additions. However, the concept of deference also fails to adequately
distinguish ritual from non-ritual. Ethical and customary norms share these same
qualities of conscious indeterminate deference, in a manner that often displaced onto
tradition and supernatural beings. These examples are arguably within the fold of
religion, but they are less arguably cases of ritual.
The distinction between reactive and objective attitudes toward agency allows us
to take a further step forward. My suggestion is that Strawson’s concept of the objective
attitude offers a more effective way to characterize those aspects of ritualized action that
Humphrey and Laidlaw analyze in terms of non-intentionality and Bloch in terms of
deference.
One of the advantages of bringing the distinction between reactive and objective
attitudes into dialogue with recent theory of ritual is that it offers richer resources for
making sense of participant attitudes toward the agency that they and others exercise. It
is important to distinguish between cases where participants see intentions linked
‘normally’ to ritualized action and those where they do not. This is so even where
scholars of religion might argue that both cases are non-intentional.
It is significant that Staal and Humphrey and Laidlaw drew on highly formalized
south Asian rituals as examples in order to conclude that issues of meaning and
intentionality are independent from ritual form. (Laidlaw and Humphrey acknowledge
that “performance-centred” rituals require the extension of their analysis of nonintentionality [2006: 282].) On the one hand, these examples offer the advantage of
rigid ritual form, facilitating the comparative analysis of this form’s relation to other
aspects of ritualized action. On the other hand, these examples occlude the importance
of ritual dynamics, of improvisation and individual variation, of choices between
alternative possible forms, of spontaneous invention, and of negotiation between ritual
participants. It is relatively easy to see what their claim that ritual is non-intentional
means with the highly formalized examples that they draw upon. It is less easy to make
sense of this claim in the case of less formalized rituals or where comparable ritual
elements are present in cases that vary in their degree of formalization. To respond that
non-intentionality is a characteristic only of formalized and pre-scripted ritual would
miss much of importance. The distinction between reactive and objective attitudes can
characterize a fuller range of types of non-intentionality.
The distinction between reactive and objective attitudes not only allows us to
further specify Humphrey and Laidlaw concept of non-intentionality and Bloch’s
concept of deference, it intersects with Houseman and Severi’s claim that ritual
performance is “an enactment of exceptional relationships,” i.e., that “the symbolic
‘work’ of the rite … consists in the establishment, on the basis of existing relations, of a
new relationship” (Houseman 2006: 420; Houseman and Severi 1998: xiv). The
distinction between reactive and objective attitudes allows us to specify one important
manner in which normal social relations are evoked or echoed but, at the same time,
transformed or displaced within ritual. It also clarifies an important dimension of ritual
virtuality according to Kapferer’s analysis. By transforming what would be a reactive
12	
  

attitude in a non-ritual context to an objective reaction in ritual, or vice versa, ritualized
action opens up a complex field for modeling alternative and variations on a key aspect
of social relations, our ascriptions of intentional agency to others and ourselves.
Indexicality, Distributed Agency, and Ritual Dynamics
This section of the article argues that the indexicality of certain ritual elements
contributes to the transformation of reactive to objective attitudes. Attempts to theorize
ritual based on models of linguistic communication have emphasized the symbolic
function of ritual elements, but this approach is limited and distracting. In asking what a
ritual “means,” we interpret its elements against the broader religious and cultural
background that gives meaning to these signs or symbols. This always leads us beyond
ritualized action itself to its alleged backdrop of signification. In semiotic terms, given
that symbols mean what they mean because of linguistic and cultural conventions, we
are led to analyze ritual as a token whose type lies beyond the ritual context itself.
Indexes are signs whose meaning is inseparable from their specific context. The word
“this,” for example, is an indexical expression, because it serves as a pointer, and, as
such, its specific meaning is inseparable from the context within which it points at
something. As a result, the concept of indexicality turns our attention toward, rather
than beyond, the ritualized actions themselves. One of the main weaknesses of the
model of ritual as linguistic communication is that is loses sight of the specificity of
ritual. The main advantage of analyzing certain ritual elements as indices is that this
focus on ‘thisness’ allows us to make better sense of ritual efficacy and dynamics.
A useful theoretical foil for clarifying the place of indexicality in ritual is Alfred
Gell’s theory of art, in his Art and Agency (1998; see Layton 2003; Bowden 2004). His
theory is based on the agency of art not on its aesthetic or communicative aspects: art is
“a system of action, intended to change the world rather than encode symbolic
propositions about it” (Gell 1998: 6). For Gell, artworks are effective mediators of
social agency because they motivate viewers to make abductions of agency: “Any
object that one encounters in the world invites the question ‘how did this thing get to be
here?’” (Gell 1998: 67). As a result, art works embody a form of “distributed agency”
that both acts on its own and points toward its originating agency. Gell’s points about
agency can be extended to other objects besides art objects: “Agency can be ascribed to
‘things’ without this giving rise to anything particularly recalling the production and
circulation of ‘art’” (Gell 1998: 23).
In Gell’s theory, the concept of agency is closely linked to the indexicality of art
objects. Here he draws on Charles Sanders Peirce’s distinction between icon, symbol
and index. An icon refers by exemplifying, exhibiting or sharing a similar structure as
its object. Architectural drawings or onomatopoeic words are icons in this sense. A
symbol refers to its object by virtue of cultural and linguistic conventions. An
interpretant (i.e., the person who interprets) understands that the word “red” refers to a
certain colour by applying a conventional rule that associates the two. The concept of
the index is more complex. It is generally spelled out in terms of a causal relation
between object and sign. Smoke is an index of fire and boot prints in the snow are an
index of the recent presence of a human being. A classic statement defines the three as
follows:
an icon is an object which takes on its sign character by reason of its
resemblance to its object; the symbol is a sign whose connection with its object
is merely associational and conventional; and the index is a sign which is really,
13	
  

dynamically, and even causally connected with what it serves to indicate (Moore
et al. 1942: 367).
The three categories of index, icon and symbol are not mutually exclusive: a given sign
can be more than one (Burks 1949; García-Carpintero 1998: 532).
The use of indexical expressions has been richly developed in two main areas of
study, analytic philosophy and ethnomethodology, with philosophers emphasizing that
the truth-value of indexicals depends on their context and ethnomethodologists argues
that their meaning does (Barnes and Law 1976: 224). They hold the basic point in
common: it is impossible to say what indexical expressions like “this is new,” “it is
raining now,” or “I am hungry” refer to unless we know the specific pragmatic context
in which the sentence was uttered (Bar-Hillel 1954). At issue here is the degree to which
it is possible to arrive at any understanding of language extracted from its context.2 The
main implication for ritual theory is this emphasis on context. Gell’s perspective is
limited, in part, because he drew to a very limited extent on these two rich sets of
literature.
Gell emphasizes the causal and non-linguistic aspects of the index, highlighting
the relevance of the concept to his proposed link between agency and art objects.
He writes that an index is a “‘natural sign’ … from which the observer can make a
causal inference of some kind, or an inference about the intentions of capabilities of
another person” (1998: 13). Jens Kreinath further specifies aspects of the index in
arguing that the concept of the index (and Gell’s work specifically) is relevant for ritual
theory, in part because it allows to makes inferences about the intentionality of ritual
participants: “It is characteristic of indexes that they not only function as vectors in
causal relations but that they are also self-referential and therefore capable of causal
inference that allows to build hypotheses about the intentions or capacities of other
persons” (2006: 468-469). On this view, four aspects of the index emerge as especially
relevant in extending Gell’s theory of art to theory of ritual: a causal relation between
object and sign; the material nature of this causal relation; the non-linguistic character
of the sign; and the particularity of this direct form of reference.
Of these four characteristics that appear to bridge Gell’s theory of art and theory
of ritual, the first three can be interpreted more broadly. This allows us to gain insights
from Gell’s work, avoiding some of the points for which he has been criticized, and to
integrate these insights with recent theory of ritual in a productive manner. That is, I
argue that it is productive to view the relation between index and object as not
necessarily causal, material or non-linguistic.
This is not to suggest that it is wrong to interpret Peirce in this triply narrow
manner. This view is one of many consistent with Peirce’s complex writings on the
matter. These and other aspects of the index are found in his fragmented and sometimes
contradictory writings on semiotics:3
• A causal connection between object and sign: “An Index is a sign which refers
to the Object that it denotes by virtue of being really affected by that Object.”
	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
   	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  
2

It is arguable that indexical representations have played a central role in the evolutionary emergence of
domain-specific cognitive mechanisms oriented to solving recurrent adaptive problems (Clarke 1996).
There is an implicit link here between the view I argue for and cognitive theories of ritual.

3

All citations are from the Commens Dictionary of Peirce’s Terms
<www.helsinki.fi/science/commens/dictionary.html>. Complete references are available there.

14	
  

• Materiality of that relation: “The index is physically connected with its
object….”
• More broadly than just a causal relation, “real connections” or “a genuine
Relation” between sign and object: “really and in its individual existence
connected with the individual object”; “a sign of its object by virtue of being
connected with it as a matter of fact.”
• A corollary of this sort of relation, contiguity: “the index, … like a pronoun
demonstrative or relative, forces the attention to the particular object intended
without describing it”; “An object, in so far as it is denoted by an index, having
thisness, and distinguishing itself from other things by its continuous identity
and forcefulness, but not by any distinguishing characters, may be called a
hecceity.”
• A quality of grabbing the attention: “A sign which denotes a thing by forcing it
upon the attention is called an index”; “an index … is a sign of its object by
virtue of … forcibly intruding upon the mind, quite regardless of its being
interpreted as a sign.”; indices “direct the attention to their objects by blind
compulsion.”
Gell’s appeal to the concept of indexicality reflects his specific concerns with
the agency of art objects. He want to frees the anthropology of art from an excessively
relativistic concern with culture, and he want to link the specific form of causality
implicit in the case of artifacts to the issue of agency. This leads him to interpret Peirce
narrowly, interpreting the index in causal, material and non-linguistic terms.
A focus on theory of ritual leads us to broaden the concept of indexicality in
each of these manners, emphasizing four themes from Peirce’s account of the index.
First, the relation between index and object can be one of contiguity (real and
existential) in addition to one of causality. A pointing finger is an index by virtue of
contiguity not cause. Second, it follows that this relation does not necessarily require a
“material imprint” (Kreinath 2006: 468). Third, the category of index includes linguistic
signs, as implied by Peirce’s reference to pronouns and developed extensively in the
subsequent philosophical literature. Fourth, it is important to take account of the
attention-grabbing quality of indices.
The following analysis draws on three specific senses of this enlarged concept of
indexicality. First, a variety of interjections and alterations in tone of voice are
characteristic of spirit possession. This type of index includes linguistic signs and is
informed by relations of contiguity more broad than those of causality.
Ethnomethodologists have noted the complex nature of this indexical function:
Interjections are primarily indexical … in that they stand for their objects by a
relationship of contiguity rather than by a relationship of convention (as in the
case of symbols) or similarity (as in the case of icons). … interjections may
index more than one object at once. In particular, they may index objects, signs,
internal states, and social relations (Kockelman 2003, 471).
For example, “wow” or “yikes” can index emotions; “hi” or “good riddance” can index
the nature of the social relationship between interlocutors; and “knock on wood” and
“God willing” can index attitudes toward agency. In spirit possession, the meaning of
certain elements of the ritual is not fleshed out in relation to some external backdrop of
cultural meaning. They mark the presence of an altered agency: here, now, in this
person, at this moment.
15	
  

Second, the messages received by mediums, and passed on in oral or written
form, and the objects that are blessed by the spirits also function as indices of
supernatural agency. Here we are close to Gell’s analysis, with the exception that these
indices are not exclusively material artifacts but also linguistic and conceptual.
To make sense of this second category of ritual indexicals, it is useful to engage
another aspect of Gell’s theory, his concept of distributed agency. In his view, art
objects are secondary agents, acting as indices of their creators agency. An agent,
according to Gell, is any “thing” (e.g. a person or object) “who is seen as initiating
causal sequences of a particular type, that is, events caused by acts of mind or will or
intention” (1998: 16; cf. 17, 19). This focus on agenct as intentional cause allows Gell
to extend the concept to include the agency of objects:
Agency is attributable to those persons (and things)…) who/which are seen as
initiating causal sequences… events caused by acts of mind or will or
intention…. An agent is the source, the origin, of causal events, independently of
the state of the physical universe (Gell 1998: 16).
This contrasts with other influential views of agency. Anthony Giddens, for
example, includes a contrafactual element that is missing in Gell’s account: “agency
concerns events of which an individual is the perpetrator, in the sense that the individual
could, at any phase in a given sequence of conduct, have acted differently” (Giddens
1984: 9). Giddens focuses on the process on agency, with its various phases of decisionmaking, intentionality, will and action in the face of potential external limitations and
constraints. Gell focuses on the endpoints of agency, on the initial cause and the result.
In his paradigmatic case, the former is inferred (or abducted) from the latter, i.e., from
the presence of material artifacts. Giddens looks at agency from the inside; Gell looks at
agency from the outsider’s perspective, analyzing cases after the fact from the agentic
signs that they present to us. For this reason, Gell’s conception is less appropriate for
the philosopher and the sociologist, but perhaps more appropriate for the anthropologist
and the student of ritual.
This sort of external approach to the concept of agency is helpful for making
sense of religious agents. That is, by focuses on what acts like an agent rather than what
thinks like an agent, we bring the concept of agency to bear on the sorts of agency that
play important roles in religious contexts. Ronald Inden argues that gods are agents
“whose very existence may be contested … [but who] may in a sense be real. … We
may take such agents to be real to the extent that complexes of discursive and
nondiscursive practices constitute and perpetuate them” (1990, 27).
William S. Sax connects this sort of distributed agency explicitly to the functions
of ritual: “Ritual is the point at which the agency distributed among other persons,
relationships, and social institutions is articulated and made manifest. … [P]ublic ritual
is precisely the point at which complex agency is articulated and confirmed” (Sax 2006:
481, 478). This point that Sax makes is central to my argument. Agency is extended,
transformed, displaced and naturalized through the ritualization of action. Specifically, I
argue that altering the relations between reactive and objective attitudes to agency are
one way in which supernatural agency is recognized or constituted. Supernatural agency
emerges through the cracks that ritual creates in our ‘normal’ perceptions of human
agency.
In this light, Gell’s concept of distributed agency draws our attention to two
points that are relevant to the indexicality of ritual. First, objects are seen as having
effects that are independent of the direct actions of their originator. It is in this sense
16	
  

that artworks have distributed agency: the agency of the artist acts secondarily, in a
mediated fashion, through these artifacts. Second, because this distributed agency stands
at a distance from its originator, we are forced to postulate the identity of that
originating agency using a process of abduction. Gell notes, “Any object that one
encounters in the world invites the question ‘how did this thing get to be here?’” (Gell
1998: 67). It is precisely at this point that altered perceptions of agency can displace the
postulate of an original human agent onto that of divine agency.
To sum up this second extension of indexicality beyond Gell, the concept of
distributed agency offers a valuable means for explaining the power of artifacts and
other material elements of ritual, as well as many of the discursive and conceptual
products of ritualized action. On the one hand, this point to the perceived affects that
ritual agents, human or supernatural, can have, effects that can extend outward into the
world more broadly through the secondary agency embodied in ritual products. On the
other hand, this very extension, from original human agent to sites of distributed
agency, creates a gap into within which divine agency can be perceived or constituted.
The third sense of an enlarged concept of indexicality that is relevant to the case
of ritual is the fact that, as Peirce noted, indices grab our attention. Gell makes a similar
point when he argues that art objects are captivating: “they are difficult to make,
difficult to ‘think’, difficult to transact. They fascinate, compel, and entrap as well as
delight the spectator” (Gell 1998: 23). However, Gell does not link this fascination
clearly to the concept of indexicality itself.
It is not a coincidence that the same is said of spirit possession: “possession
continues to hold the anthropological gaze despite heroic attempts to tame it, render it
harmless or understood” (Boddy 1994: 407). Vincent Crapanzano underlines this
compelling quality of spirit possession: “‘Fascination’ captures the magic, the demonic
power, of possession itself…” (Crapanzo 2006: 200-201). The parallel between the
fascination of the art object and the fascination of possession is that both force us to ask
questions about the agency that lies behind what we see before us. I argue that
possession is fascinating in large part due to its superposition and transformation of
reactive and objective attitudes. We see someone, perhaps a neighbor or relative, in
action, but we are forced, by the ritualized nature of that action, to alter or our attitude
toward this manifestation of human agency.
This transformation of the balance between these two attitudes to agency plays a
role in the constitution of the sacred. Reactive attitudes are displaced in ritual: we react
naturally, but this reaction cannot be attached to the ritual actor, toward whom we adopt
an objective attitude. The postulated supernatural agents often serve as the objects of
these displaced reactive attitudes. Where Houseman and Severi draw attention to
exceptional relations, I argue that it is our exception reactions, in part constituted by
these incommensurable social relations, which are more significant. Ritual invokes a
disjunction, as our attitudes to agency are evoked but displaced.
To sum up, Gell proposes two useful concepts: indexicality and distributed
agency. Making sense of ritual requires interpreting indexicality in a broader sense than
Gell does: including linguistic phenomena, such as interjections; recognizing that divine
agency can emerge from the gap between original human agents and sites of distributed
agency; and the fascination that is evoked, in part, by the perceived disjunction that is
produced in our perceptions of agency when it is transformed, displaced, or distributed.
One of Gell’s goals in invoking the concept of the index was to distance the
anthropology of art from its emphasis on cultural context. According to Peirce, the
symbol can only be read by paying attention to the broad cultural conventions that
constitute its meaning, but the index has a real, existential link to a specific context.
17	
  

However, extending the concept of the index beyond a narrow view focused on causal
and material relations brings culture back into the equation. For example, interjections
are indexicals, but part of their meaning is governed by conventions: a “wow!” by any
other name is not necessarily a “wow!” Gell’s view of indices is simply untenable in its
radical departure from culture and convention. Ethnomethodologists classify indices as
“deictic tokens,” and analyzing the cultural context of these context-driven signs is one
of their stocks in trade: “The anchoring of deictic tokens to a specific time and place
relies on cultural expectations that organize their salience” (Duranti 1997: 352).
The final two sections of this article explores a specific set of indexicals, signs of
altered agency in Brazilian spirit possession rituals, and it argues that the work of these
ritual elements reflects a very specific set of cultural expectations. This is intended to
accomplish the two overarching goals of this contribution to ritual theory: to set out
concepts that offer a broad applicability in other contexts, while making it possible to
take account of specific social and cultural contexts.
Virtuality, indexicality and attitudes to agency in Kardecism and Umbanda
This final section illustrates the value of the theoretical approach outlined above by
analyzing spirit possession in Kardecism and Umbanda. From the perspective of ritual
theory, the key task set out by this brief overview of two Brazilian religions is to explain
the ritual differences. A further challenge is to propose a theoretical frame that draws on
the strengths of recent important work in ritual theory.
The distinction between reactive and objective attitudes offers important insights
into spirit possession, in part because it highlights the need to analyze the details of
possession in specific religious and social contexts. As Houseman and Severi suggest,
ritualized action constitutes special relationships. Here I offer a specific account of what
this means in this particular ritual context: the pervasive patron-client relationships of
Brazilian society are echoed and transformed in these rituals. Specifically, I argue that
the relation between reactive and objective attitudes is altered in these contexts. The
sacralizing function of a shift between reactive and objective attitudes constitutes a key
dimension of the rituals’ virtuality, of their capacity to allow participants to re-imagine,
redirect, and reorient their capacity for agency with respect to the everyday
circumstances of life.
Possession rituals transform attitudes toward agency in three ways. (Recall that a
reactive attitude reflects a judgment that others’ agency is normal, above all, intentional;
an objective attitude reflects a judgment that their actions are not entirely their own.)
First, an objective attitude toward the medium who is possessed replaces the reactive
attitude that holds toward this member of one’s community outside the ritual context.
This shift from reactive to objective attitudes is based on the non-intentionality of the
medium’s agency and it is marked by certain indexicals. Second, as a corollary, reactive
attitudes are displaced onto a supernatural patron, the possessing spirit, who acts more
generously and less self-interestedly than is the case with the normal patron-client
relationship. The messages that are conveyed by mediums, both oral and written, and
the objects blessed by the spirits and later circulated function as indices of that altered
agency. At this point, Gell’s analysis is particularly appropriate. The particular tension
between reactive and objective attitudes here depends on the specific character of
Brazilian society, a reminder of the need for ritual theory to take social and cultural
contexts into account. Third, the clients’ attitude toward their own actions, during and
as a direct result of the ritual, shifts toward the objective. This shift from reactive to
objective attitudes is based on the normative constraints that inhere in the ritual process.
18	
  

In all three cases, the indexicality of certain ritual elements is essential to unhooking
reactive responses like gratitude from the reciprocity that governs non-ritual social
relations. In general, the virtuality of spirit possession in Kardecism and Umbanda
models a more functional set of patronage relations that are leveraged free from
constraining links of reciprocity.
Various aspects of possession rituals in Umbanda and Kardecism illustrate the
role of indexicals in prompting a shift toward objective attitudes regarding the medium:
dramatic gestures, sudden movements, eye rolling, and sharp cries mark the moment of
possession, more so toward the Afro end of Umbanda; the spirits’ ‘horses’ often have a
distinctive rolling gait in rituals at the Afro-Brazilian end of Umbanda’s spectrum of
ritual forms; changes in tone and pacing of voice mark possession; more specifically,
guttural interjections are characteristic of the manner in which possessed mediums
speak in Umbanda; distinctions between these guttural interjections gives evidence of
different types of spirits, e.g., caboclos and pretos velhos; the hushed and attentive
response of non-participants/clients also marks specific points in the ritual that are
characterized by the presence/agency of spirits.
These signs are all indexical in that they are caused by or contiguous with the
possession process. As such they point to an alteration in the agency of the medium.
That is, these indexicals are the basis on which the non-participants/clients make
judgments about the nature of agency manifested by the medium, informing a shift of
attitude from reactive to objective. The judgment results in the medium having the
status of an artwork, in terms of Gell’s analysis, i.e., an object, a site of distributed
agency, that forces us to search for an originary agency behind it. The postulate of a
supernatural agent follows: clients face what appears to be an agent yet have strong
reasons to believe that the person before them is not acting normally; the obvious
conclusion is that an occluded agent is acting through them. What this article adds is a
clarification of the relation between two attitudes to agency that inform this reaction.
As noted above, Gell’s attempt to sidestep the cultural contingency that shapes
the interpretation of indexicals fails. I suggest that the prominence of spirit possession
in Brazil is a function of the fact that the general function of ritual that I point to,
altering relations between attitudes to agency, resonates more strongly with attitudes to
agency that are common within Brazilian society as a whole.

19	
  

	
  

Objective
attitude
	
  

Reactive
attitude
Normative
Constraints
Non-intentionality

	
  

Figure 2a. Objective attitudes to agency as a function of
non-intentionality and normative constraints
(Canada/U.S.A.).	
  
	
  
	
  

Reactive
attitude

Objective
attitude

Normative
Constraints
Non-intentionality

Figure 2b. Objective attitudes to agency as a function of
non-intentionality and normative constraints (Brazil).
The threshold at which reactive attitudes shift to objective attitudes differs
between cultures (Figures 2a and 2b). In Brazil, the scope of one’s agency in the nondomestic sphere is more limited than is generally the case, for example, in North
America or Western Europe. As a result, objective attitudes are more often appropriate.
Along the “non-intentionality” axis in Figures 2a and 2b, the popular sense grants a
great scope of non-intentionality; the point at which one is no longer responsible for
what has happened. This reflects the fact that bureaucratic and other public systems are
inefficient, irrational and, on occasion, corrupt, leading to a correlated reliance on
personal relations (Buarque de Holanda, 1999: 139-151; Queiroz 1976: 33-159;
20	
  

DaMatta 1983).4 It also reflects a correlated belief that a variety of spirits obstruct one’s
efforts. Belief in spirit possession is widely held, especially among the lower classes.
Encostos are believed to be very common. These are mildly harmful spirits that
obstruct, distract, hinder, cause illnesses, etc. (interpreted in Kardecism as mild cases of
obsession). Self-help programs on Brazilian television discuss encostos where Oprah
and Dr. Phil would be discussing dysfunctional self-image. Along the “normative
constraints” axis in Figures 2a and 2b, the reverse is true. Because confidence in
impersonal system is justifiably low, those who would be perceived, in Canada and the
USA, as normatively constrained are, in Brazil, more often seen as having the latitude,
if they so desire, to dar um jeito (“find a way”) (Barbosa 2006). As a result, the ritual
shift in attitudes toward agency comes easier. It reflects the relational tensions between
reactive and objective attitudes in Brazilian society more closely. In Brazil, it is easier
and more ‘natural’ to adopt an objective attitude toward a medium, displacing one’s
reactive attitude onto a postulated supernatural being.
This perspective allows us to take one step toward making sense of a key ritual
difference between Kardecism and Umbanda. The two religions exhibit important
differences in the social make-up of practitioners: Kardecists are more likely white,
middle-class, and well educated; Umbandists more likely to be pardo or black, poor,
and poorly educated. That is, Kardecists have, on average, significantly more social and
economic capital within Brazilian society. Within Umbanda, with its spectrum of ritual
forms, a homologus tensions exists between “white” and Afro forms (Engler
forthcoming). The fact that the rituals in Umbanda at the “white” end of the spectrum
more closely resemble Kardecism is correlated with higher middle-class and white
participation. In terms of the theoretical frame proposed here, practitioners of the two
religions are rehearsing and reorienting different tensions between reactive and
objective attitudes toward agency, correlated with their ability to wield social capital.
That is, the virtuality of rituals rehearse and model different social landscapes in the two
cases.
A key ritual difference is correlated with this social difference: Kardecist
mediums remain conscious of and largely in control of the presence of the spirits they
receive; Umbandist mediums enter an unconscious trance state and are more likely to
require subsidiary rituals to free them from the spirit that has possessed them. Because
of their class and status locations within Brazilian society, Kardecists are more like
North Americans in terms of the culturally predicated attitudes toward agency that they
bring to the ritual table. The less radical displacement of reactive attitudes reflect the
fact that Kardecists are less likely to be accustomed to adopting an objective attitude
toward their own agency.
Conclusion
Agency is constrained or facilitated by our attitudes toward it, and these vary across
cultures, classes, and other social boundaries. As Gell notes, agency is “a culturally
prescribed framework for thinking about causation, when what happens is (in some
vague sense) supposed to be intended in advance by some person-agent or thing-agent”
(Gell 1998: 17). In this light, a focus on the virtuality of ritual—its creation of spaces
	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
   	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  
4

Of course, Brazil’s impersonal systems are inefficient and its personal systems efficient only by
comparison, generally to Western Europe and English-speaking North America. It is important not to
jump to the conclusion that these factors are necessarily correlated with, much less the cause of, a
‘backward’ or ‘deficient’ modernity in Brazil (Souza 1999).

21	
  

for rehearsing alternative attitudes toward agency—offers a theoretical framework for
analyzing ritualized action in a manner responsive to cultural specificity.
The elements of a theoretical approach sketched here are not meant to be
complete or exclusive. The indexicality of ritual (or formal aspects of ritual more
generally) are not determinative or solely responsible for shaping people conceptions of
and attitudes to agency. Nor are reactive and objective attitudes to agency exclusive of
or more important than a variety of other factors. For example, the notion of person in
Kardecism is distinct from that of Umbanda in its emphasis on rational agency and the
modification of individual behaviour (Cavalcanti 1983). This religious anthropology
might well play a greater role in shaping attitudes to agency that the elements of ritual
form that I have focused upon.
It might seem overly functionalist or reductionist to suggest that these rituals can
frame functional and even strategic displacements of intentionality. This would appear
to fly in the face of Kapferer’s denial that ritual is “a coded symbolic formation whose
interpretation or meaning is ultimately reducible to the sociopolitical and psychological
world outside the ritual context” (Kapferer 2004: 46; cf. 2002: 118). Kapferer’s
insistence on the automony of ritual takes a healthy step back from overly reductionist
approaches. At the same time, he is often too stark in arguing for the independence of
ritual from the social field of non-ritual: “the virtual of ritual is a thoroughgoing reality
of its own, neither a simulacrum of realities external to ritual nor an alternative reality”
(Kapferer 2004: 37). His core insight, that rituals allow participants to “reimagine (and
redirect or reorient themselves) in the everyday circumstances of life” would make no
sense if the space created by the ritual bore no relation to the non-ritual sphere
(Kapferer 2006: 673-674). It is the dynamic tension between like and unlike that allows
the virtuality of ritual “to realize human constructive agency” (Kapferer 2004: 47).
The approach I sketch could be used as part of narrowly (and indefensibly)
functionalist or reductionist arguments. I point rather to correlations between attitudes
to agency in the ritual sphere and non-ritual spheres, suggest that elements of ritual form
contribute to this relation of like and un-like, and argue that these correlations vary to
some extent by culture and class. Brazilian spirit-possession rituals do not simply echo
or reflect broader cultural factors; they also allow ritual participants to reimagine and
rehearse dynamic conceptions of and attitudes to the agency of themselves and others;
and this has implications outside of the ritual context. The indexicality of rituals, in part,
both influences and reflects changes in attitudes to agency; and this constitutes an
important element of the virtuality of the rituals. Given their virtual framing of attitudes
to agency and intentionality, rituals could potentially do several sorts of work for
individuals and groups: adapting or dislocating participants from their broader social
context; preparing them for more effective non-ritual agency or compensating for its
lack; reflecting, distorting, critiquing, romanticizing social reality. The extent to which
any given set of rituals performs these or other functions is a largely empirical issue,
one not addressed here. It would be one-sided and limiting to analyze rituals solely in
terms of their offering psychological compensation to those whose social position
leaves them with hampered agency in ‘real’ life: in terms of the present analysis, by
shifting participants’ views of structural constraints from dysfunctionally reactive to
functionally objective, reorienting attitudes to agency in relation to structural norms.
This may well be an important part of the story, but other possibilities should also be
considered: e.g., that religious rituals can constitute a social space within which altered
attitudes to agency are correlated with distinct status markers.
In substantive terms, I argue that Brazilian spirit-possession rituals create a
virtual space where participants work out alternative senses of agency and
22	
  

intentionality, and these latter bear certain relations to the broader field of Brazilian
social relations. The distinct attitudes to agency that are prevalent in Brazilian society
make spirit possession more effective in two senses of that word. First, the rituals
succeed in evoking or constituting the sacred more effectively because their
displacement of reactive attitudes comes more easily to those who are more used to
making this mental move. Second, the rituals offer a space for the rehearsal of a type of
attitude adjustment that is functional in Brazilian society in a way that it is not in
Canadian or American society: it is useful to recognize when others are not able to act
fully as agents and to displace one’s reactive attitudes toward figures that hold positions
that allow them to function as patrons.
My theoretical claim is a limited one: that certain formal aspects of ritual offer
cues that help constitute a virtual space for participants, within which culturally-specific
relations between reactive and objective attitudes can be modified. Cultural and socioeconomic factors, as well as religious conceptions of the self as agent, inform the
relevant aspects of ritual dynamics. This basic approach could be extended to analyze
other dimensions of relations between ritual form and conceptions of agency and
intentionality.
Bibliography
Aubrée, Marion & François Laplantine (1990). La table, le livre et les Esprits:
Naissance, evolution et actualité du movement social spirite entre France et Brésil.
Paris: Éditions Jean-Claude Lattès.
Bar-Hillel, Yehoshua (1954). Indexical expressions. Mind (N.S.) 63(251): 359-379.
Barbosa, Lívia (2006 [1992]). O jeitinho brasileiro: a arte de ser mais igual do que os
outros. Rio de Janeiro: Elsevier/Editora Campus.
Barnes, Barry & John Law (1976). Whatever should be done with indexical
expressions? Theory and Society 3(2): 223-237.
Bastide, Roger (1967). Le spiritisme au Brésil. Archives de Sciences Sociales des
Religions 24(1): 3-16
Birman, Patricia (1982). Laços que nos unem: ritual, família e poder na Umbanda.
Religião e Sociedade 8: 21-28.
—— (1985). Identidade social e homosexualismo no Candomblé. Religião e Sociedade
12(1): 2-21.
—— (1995). Fazer estilo criando gêneros. Possessão e diferenças de gênero em
terreiros de Umbanda e Candomblé no Rio de Janeiro. Rio de Janeiro: Editora
UERJ/Relume Dumará.
Bloch, Maurice (2006). Deference. In Jens Kreinath, Jan Snoek, & Michael Stausberg
(eds.) Theorizing Rituals: Issues, Topics, Approaches, Concepts, 495-506. Leiden and
Boston: Brill.

23	
  

Boddy, Janice (1994). Spirit possession revisited: beyond instrumentality. Annual
Review of Anthropology 23: 407-434.
Bowden, Ross (2004). A critique of Alfred Gell on art and agency. Oceania 74(4): 309324.
Boyer, Véronique (1992). De la campagne à la ville: la migration du caboclo. Cahiers
d’Études africaines 32(125): 109-127.
Brown, Diana (1977). Umbanda e classes sociais. Religião e Sociedade 1.
——— (1985). Uma história da umbanda no Rio. In Umbanda e Política, 9-42. Rio de
Janeiro: Marco Zero.
——— (1994 [1986]). Umbanda: Religion and Politics in Urban Brazil. New York:
Columbia University Press.
Brown, Diana De G. & Mario Bick (1987). Religion, class, and context: Continuities
and discontinuities in Brazilian Umbanda. American Ethnologist 14(1): 73-93.
Brumana, Fernando Giobellina & Elda Gonzales Martinez (1989). Spirits from the
Margins: Umbanda in São Paulo. Uppsala: Uppsala University Press.
Buarque de Holanda, Sérgio (1999 [1936]). Raízes do Brasil. São Paulo: Companhia
das Letras.
Burks, Arthur W (1949). Icon, index, and symbol. Philosophy and Phenomenological
Research 9(4): 673-689.
Caciola Nancy (2000). Mystics, demoniacs, and the physiology of spirit possession in
medieval Europe. Comparative Studies in Society and History 42(2): 268-306.
Camargo, Candido Procôpio Ferreira de (1961). Kardecismo e Umbanda: uma
interpretação sociológica. São Paulo: Pioneira.
Carneiro, Edison (1977 [1948]). Candomblés da Bahia. 5th edition. Rio de Janeiro:
Editora Civilização Brasileira.
Cavalcanti, Maria Laura Viveiros de Castro (1983). O mundo invisível: cosmologia,
sistema ritual e noção da pessoa no Espiritismo. Rio de Janeiro: Zahar.
Chesnut, R. Andrew (2003). Pragmatic consumers and practical products: The success
of pneumacentric religion among women in Latin America’s new religious economy.
Review of Religious Research 45(1): 20-31.
Clarke, Murray (1996). Darwinian algorithms and indexical representation. Philosophy
of Science 63(1): 27-48.

24	
  

Concone, Maria Helena Villas Bôas (2001). Caboclos e pretos-velhos da Umbanda. In
Reginaldo Prandi (ed.) Encantaria brasileira: o livro dos mestres, caboclos e
encantados, 281-303. Rio de Janeiro: Pallas.
Crapanzano, Vincent (2006). Postface. Culture and Religion 7(2): 199-203.
DaMatta, Roberto (1983 [1978]). Carnavais, malandros e heróis: para uma sociologia
do dilema brasileiro. 5th edition. Rio de Janeiro: Zahar. [Carnivals, Rogues, and
Heroes: An Interpretation of the Brazilian Dilemma, Notre Dame, IN: University of
Notre Dame Press, 1991.]
——— (1987). Relativizando: Uma introdução à antropologia social. Rio de Janeiro:
Rocco.
Duranti, Alessandro (1997). Indexical speech across Samoan communities. American
Anthropologist (N.S.) 99(2): 342-354.
Engler, Steven (2008.) Field Guides to the Study of Religion. Religious Studies Review
34(1): 17-29.
——— (Forthcoming). Umbanda and Hybridity. Numen.
Fischer, Brodwyn (2004). Quase pretos de tão pobres? Race and social discrimination in
Rio de Janeiro’s twentieth-century criminal courts. Latin American Research Review
39(1): 31-59.
Fry, Peter (1978). Manchester e São Paulo: industrialização e religiosidade popular.
Religião e Sociedade 3: 25-52. Reprinted in Para inglês ver, 54-86. Rio de Janeiro:
Zahar (1982). English version in John D. Wirth & Robert Jones (Eds.) Manchester and
São Paulo: Problems of Rapid Urban Growth. Stanford: Stanford University Press
(1978).
——— (1982). Homosexualidade masculina e cultos afro-brasileiros. In Para inglês
ver: identidade e política na cultura brasileira, 54-86. Rio de Janeiro: Zahar.
García-Carpintero, Manuel (1998). Indexicals as token-reflexives. Mind (N.S.)
107(427): 529-563.
Gell, Alfred (1998). Art and Agency: An Anthropological Theory. Oxford: Clarendon
Press.
Giddens, Anthony (1984). The Constitution of Society: Outline of the Theory of
Structuration. Berkeley: University of California Press.
Harding, Rachel E. (2005). Afro-Brazilian Religions. In Lindsay Jones (ed.),
Encyclopedia of Religion, 2nd ed., 119-125. New York: Macmillan Reference.

25	
  

Hayes, Kelly E. (2006). Caught in the crossfire: Considering the limits of spirit
possession: A Brazilian case study. Culture and Religion 7(2): 155-175.
Hess, David J. (1991). Spirits and Scientists: Ideology, Spiritism, and Brazilian Culture.
University Park: Pennsylvania State University Press.
Hoffman Kelly & Miguel Angel Centeno (2003). The lopsided continent: Inequality in
Latin America. Annual Review of Sociology 29: 363-90.
Houseman, Michael (2006). Relationality. In Jens Kreinath, Jan Snoek, & Michael
Stausberg (eds.) Theorizing Rituals: Issues, Topics, Approaches, Concepts, 413-428.
Leiden and Boston: Brill.
Houseman, Michael & Carlo Severi (1998 [1994]). Naven or the Other Self: A
Relational Approach to Ritual Action. Leiden and Boston: Brill.
Humphrey, Caroline & James Laidlaw (1994). The Archetypal Actions of Ritual: A
Theory of Ritual Illustrated by the Jain Rite of Worship. Oxford: Oxford University
Press.
Inden, Ronald B (1990). Imagining India. Oxford: Blackwell.
Jacob, Cesar Romero, Dora Rodrigues Hees, Philippe Waniez & Violette Brustlein
(2003). Atlas da filiação religiosa e indicadores sociais no Brasil. São Paulo: Edições
Loyola.
Kapferer, Bruce (2002). Sorcery, modernity and the constitutive imaginary: Hybridising
continuities. Social Analysis 46(3): 105-128.
——— (2004). Ritual dynamics and virtual practice: Beyond representation and
meaning. Social Analysis 48(2): 35-54.
——— (2006). Virtuality. In Jens Kreinath, Jan Snoek, & Michael Stausberg (eds.)
Theorizing Rituals: Issues, Topics, Approaches, Concepts, 671-684. Leiden and Boston:
Brill.
Kockelman, Paul (2003). The meanings of interjections in Q’eqchi’ Maya: From
emotive reaction to social and discursive action. Current Anthropology 44(4): 467-490.
Kreinath, Jens (2006). Semiotics. In Jens Kreinath, Jan Snoek, & Michael Stausberg
(eds.) Theorizing Rituals: Issues, Topics, Approaches, Concepts, 429-470. Leiden and
Boston: Brill.
Kreinath, Jens, Jan Snoek, & Michael Stausberg (2006). Ritual Studies, Ritual Theory,
Theorizing Rituals--An Introductory Essay. In Jens Kreinath, Jan Snoek, & Michael
Stausberg (eds.) Theorizing Rituals: Issues, Topics, Approaches, Concepts, xiii-xxv.
Leiden and Boston: Brill.

26	
  

Laidlaw, James & Caroline Humphrey (2006). Action. In Jens Kreinath, Jan Snoek, &
Michael Stausberg (eds.) Theorizing Rituals: Issues, Topics, Approaches, Concepts,
265-283. Leiden and Boston: Brill.
Landes, Ruth (1947). The City of Women. New York: MacMillan.
Layton, Robert (2003). Art and Agency: A reassessment. Journal of the Royal
Anthropological Institute (N.S.) 9(3): 447-464.
Lewgoy, Bernardo (2006). Incluídos e letrados: reflexões sobre a vitalidade do
espiritismo kardecista no Brasil atual. In Faustino Teixeira and Renata Menezes (eds.)
As religiões no Brasil: continuidades e rupturas, 173-188. Petrópolis: Vozes.
Lovell, Peggy & Charles Wood (1998). Skin color, racial identity, chances in Brazil.
Latin American Perspectives 25(3): 90-109.
Machado, Ubiratan (1983). Os intelectuais e o espiritismo. Rio de Janeiro: Antares.
Monroe, John L. (2008). Laboratories of Faith: Mesmerism, Spiritism and Occultism in
Modern France. Ithaca and London, Cornell University Press.
Moore, Willis, Gustave Bergmann & Ray H. Dotterer (1942). The indexical and the
presentative functions of signs. Philosophy of Science 9(4): 367-375.
Motta de Oliveira, José Henrique. (2007). Eis que o caboclo veio à Terra ‘anunciar’ a
Umbanda. História, imagem e narrativas 4(2): 176-188.

Natividade, Marcelo & Leandro de Oliveira (2007). Religião e intolerância à
homosexualidade: tendências contemporâneas no Brasil. In Vagner Gonçalves da Silva
(ed.), Intolerância religiosa: impactos do neopentecostalismo no campo religioso afrobrasileiro, 261-302. São Paulo: EdUSP.
Ortiz, Renato (1999 [1978]). A morte branca do feiticeiro negro: Umbanda e sociedade
brasileira. 2nd edition. Petrópolis: Vozes.
Pierucci, Antônio Flávio & Reginaldo Prandi. 2000 Religious diversity in Brazil:
Numbers and perspectives in a sociological evaluation. International Sociology 15(4):
629-639.
Prandi, Reginaldo (2005). Segredos guardados: orixás na alma brasileira. São Paulo:
Companhia das Letras.
Queiroz, Maria Isaura Pereira de (1976). O mandonismo local na vida política
brasileira e outros ensaios. São Paulo: Editora Alfa-Omega.
Rasmussen, Susan J. (1994). The ‘head dance’, contested self, and art as a balancing act
in Tuareg spirit possession. Africa 64(1): 74-98.
27	
  

Santos, Jocélio Teles dos. 1992. La divinité caboclo dans le candomblé de Bahia.
Cahiers d’Études africaines 32(125): 83-107.
——— 1995. O dono da terra: o caboclo nos candomblés da Bahia. Salvador: Editora
SarahLetras.
Sax, William S (2006). Agency. In Jens Kreinath, Jan Snoek, & Michael Stausberg
(eds.) Theorizing Rituals: Issues, Topics, Approaches, Concepts, 473-481. Leiden and
Boston: Brill.
Severi, Carlo (2006). Language. In Jens Kreinath, Jan Snoek, & Michael Stausberg
(eds.) Theorizing Rituals: Issues, Topics, Approaches, Concepts, 583-593. Leiden and
Boston: Brill.
Sharp, Lynn L (2006). Secular Spirituality: Reincarnation and Spiritism in NineteenthCentury France. Lanham, MD: Lexington Books.
Silva, Fábio Luiz da (2005). Espiritismo: história e poder (1938-1949). Londrina, PR:
EDUEL.
Silva, Vagner Gonçalves da (2005 [2000]). Candomblé e Umbanda: caminhos da
devoção brasileira, 2nd ed. São Paulo: Selo Negro Edições.
Souza, Jessé de, Ed (1999). O malandro e o protestante: a tese weberiana e a
singularidade cultural brasileira. Brasília: Editora Universidade de Brasília.
Staal, Fritz (1979). The meaninglessness of ritual. Numen 26: 2-22.
Stoll, Sandra Jacqueline (1999). Entre dois mundos: o Espiritismo da França e no
Brasil. Unpublished doctoral dissertation in anthropology. Universidade de São Paulo.
Strawson, Peter F. (1976). Freedom and resentment. In Freedom and Resentment and
Other Essays London: Methuen. 1-25.
Warren Jr., Donald (1968). Spiritism in Brazil. Journal of Inter-American Studies 10(3):
393-405.
Xavier, [Chico] Francisco Cândido [spirit André Luiz] (2006a [1944]). Nosso Lar. Rio
de Janeiro: Federação Espírita Brasileira.
——— [spirit Humberto de Campos] (2006b [1938]). Brasil, coração do mundo, pátria
do evangelho. Rio de Janeiro: Federação Espírita Brasileira.
Yearley, Steven (1985). Vocabularies of freedom and resentment: A Strawsonian
perspective on the nature of argumentation in science and the law. Social Studies of
Science 15(1): 99-126.

28