Human Studies

, Volume 33, Issue 2, pp 221–227

What is Special About Body Based Reference Frame?


    • Indian Institute of Information Technology
Research Paper

DOI: 10.1007/s10746-010-9161-x

Cite this article as:
Khetrapal, N. Hum Stud (2010) 33: 221. doi:10.1007/s10746-010-9161-x


Classifying spatial frames of references have placed egocentric/body-based representations on muddy grounds. The traditional taxonomy places it under the deictic distinction while the Levinson’s terminology does not provide a special status for it but classifies it along with the relative frame of reference. Research from other areas of cognition has come up with other implied classifications that are motivated by the special role played by these egocentric representation(s). Tangled among such issues is the fuzzy distinction between egocentric and body based representations. The current paper takes up exactly this issue and proposes to sub classify egocentric representations into two different subtypes namely the first- and the second-order representations. The proposed distinction serves an essential purpose for understanding important cognitive processes like spatial transformation, mental perspective taking, and so on.


Spatial cognitionEgocentric representationBodyViewpoints

Types of Representations for Spatial Cognition

Researchers have often distinguished between two different types of representations required for spatial cognitive ability and navigation namely, object based representations and egocentric representations (see Mou et al. 2004; Wang and Spelke 2000). The egocentric representations specify locations with respect to navigators/observers/viewers while the object-based representations are important for coding spatial relations of object(s) with respect to other object(s), though some researchers still maintain that these representations could be aligned with viewer’s perspective1 (Mou and McNamara 2002). Thus, the object based representations are considered to be mediated by an intrinsic frame of reference usually involving situations where objects (located objects) are located with respect to the inherent features of the chosen reference object. Levinson (2003) labels this reference frame as intrinsic in nature. According to him, the intrinsic system comprises of binary relations where just two elements are required to determine the reference and thus are consequently quite simple in nature. These relations are unable to support a consistent or sustained spatial relations/inference. Levinson’s intrinsic frame of reference could serve as an important basis for mentally rotating objects. The argument structure of a spatial description is the defining characteristic of a reference frame in Levinson terminology (1996, 2003). He offers the relative reference frame classification for a ternary description involving located objects, reference object, and viewpoints.

Levinson’s terminology stands in contrast to the descriptions offered under the framework of the traditional taxonomy (see Carlson-Radvansky and Irwin 1993; refer Table 1). According to the traditional taxonomy, reference frame distinctions are dependent upon the features of the chosen reference object (people or objects). Spatial descriptions involving either a speaker or an addressee are classified as instances of deictic or egocentric reference frames. On the other hand, spatial descriptions that involve objects are classified as instances of intrinsic reference frames. Descriptions like, “a toy-ball is to the left of the toy-car” and “a toy-car is in front of me” are both considered examples of deictic or egocentric reference frames (traditional taxonomy). While Levinson’s terminology considers only the description, “a toy-car is in front of me” as an example of intrinsic reference frame as it entails just two elements. Thus, there appears to be no special room for egocentric reference frame in his scheme of classification. However, “a toy-ball is in front of the toy-car” is considered an example of intrinsic reference frame under both the classifications but for different reasons.
Table 1

Traditional vs. Levinson’s Taxonomy

Spatial descriptions

“A toy-ball is to the left of the toy-car”

“A toy-car is in front of me”

“A toy-ball is in front of the toy-car”









Problems with Classificatory Schemes

The two classificatory schemes are not consistent with other findings in the literature which show efficient processing of egocentric transformations over object-based transformations (Wraga et al. 2000; Amorim et al. 2006). Efficient processing of egocentric transformations is hypothesized to be sub-served by holistic transformations of bodies as opposed to piecemeal transformations of objects (Amorim et al. 2006). Furthermore, egocentric transformations play an important role in both basic (e.g. spatial localization) and complex computations (e.g. mental rotation and perspective taking; for technical application example, see Khetrapal 2010). The current paper thus attempts to distinguish between object-based and egocentric representations and further sub-classify the egocentric representations into first and second order (refer Table 2).
Table 2

New Taxonomy

Spatial descriptions

“A toy-ball is to the left of the toy-car”

“A toy-car is in front of me”

“A toy-ball is in front of the toy-car”


Egocentric: second order

Egocentric: first order

Intrinsic/object based

Self-Based/Body-Based or Egocentric Representations

Self-based or rather body-based representations are required for relating objects spatially to one’s body. In this state of affair, the axes of one’s body (front–back, left–right, and up–down) play an important role. This first order egocentric representation should be contrasted with the position adopted by Lakoff and Johnson (1999), who explained that the bodily reference frames could be projected onto another object which as a result could be defined as having bodily axes (front, back and so on). But such a projection leads to ambiguity in interpretation in statements such as the “doll lies in front of the ball”. How should this be interpreted when a circular entity like ball has no intrinsic fronts? Contrast this to a situation where the speaker says that the “doll lies to the front of the car”? It becomes easier for the listener to assign a “front” to the car which corresponds to the frontal direction of a moving car. But is this assignment based on the fact that the car really has a front or the bodily axes of the driver is usually projected onto the car for the sake of convenience? It is precisely because of these questions and considerations that I treat the first order egocentric representations as a different category. I would consider that the speaker adopts an object-based representation when he or she localizes the doll in front of the car. An egocentric perspective would have been adopted if the doll lies to the front of the speaker and the egocentric and the object-based reference frames correspond in case the doll lies in between the speaker and the car.

Why should the distinction between two basic types, that is, object-based representations and self-based representations motivate a write up? This distinction could be considered clearer by taking the example cited by Kozhevnikov et al. (2006), “one could imagine standing near the northern side of a map lying on a table and then imagine the map rotating on the table 180° until the southern side is closest to the oneself, or one could imagine oneself moving 180° around the table to view the map from the southern side”. As previously mentioned, several studies have also reported efficient processing of egocentric transformations over object-based transformations (Wraga et al. 2000; Amorim et al. 2006). Could this benefit accrue from the special status of our bodies or embodied cognition? Some researchers explain that this might actually be the case (see Amorim et al. 2006).

For present concerns, the viewer’s perspective (originating from the viewer’s body) will be treated as part of body-based representations. In Levinson’s terminology, egocentric representation does not have a separate special status and is usually classed with the relative frame of reference that comprise of ternary relationships because spatial relations among objects are determined by three objects (e.g. “the toy is behind the chair” from the perspective of the speaker of this sentence). Here comes the important specification and for present terminology self-based, body-based, and egocentric representations are considered different names of same representations. At the same time, these are considered distinctly from other means of spatial localizations where objects are used as reference. This is because egocentric representations are considered the first step towards embodied cognition. Thus on logical grounds, egocentric representations could be defined as comprising of two different subtypes that is, under some situations observers use their own body to spatially localize objects giving way to simple representations (e.g. “this object lies to my left”). Or, in other cases, the body is used as an origin but the observers could be inclined to locate an object (the located object) to the left of a chosen reference object even though the located object might be in reality located to the front or the back of the reference object. These could be henceforth called the second-order egocentric representations.

The viewing perspective of the observer plays an important role in the second but not in the first case. This occurs for the reason that differences in viewing point will involve aspects of perspective transformations (also be termed egocentric perspective transformation) that helps to update one’s own location or orientation. This is an important advance as humans do not remain static at one place but are quite mobile and hence we constantly need to keep a stock of ourselves with respect to our changing environments and/or changes that occur as a result of our own movements either in mental or physical space.

Localizing objects in a static place should be dependent upon rather basic representations and processes, for instance while looking onto a table top it is quite easy to say that “a toy ball lies to the left of the toy car” even though the ball might be in reality to the front of the toy car. Therefore, adding viewing perspective to simple egocentric representations should yield rather different computations and hence I consider it as a different subtype. On the other hand, updating of spatial relations in a mobile environment cannot be dependent upon static representations for instance; studies have shown that subjects take longer to decide what a scene would look like if the new imagined position is misaligned with the actual position of the subject (Amorim and Stucchi 1997; Creem et al. 2001).

There are other instances of egocentric perspective transformations where the observer needs to adopt the perspective of another human. Such kind of abilities could be important for mental perspective taking and empathic responding (Thakkar et al. 2009). There have been others who also provide support for the mediating role of spatial perspective taking when adopting the viewpoint of another human where the mental rotation of the ‘self’ plays the most important role (Graf 1994). Kozhevnikov and Hegarty (2001) and Kozhevnikov et al. (2006) further suggest that spatial perspective taking operate within an egocentric reference frame. In my theoretical development, the second order egocentric representations play an important role in this regard as well. Here I wish to set my position apart from those who consider adopting another’s perspective as a disembodied process akin to imagining oneself out of one’s own body (Blanke et al. 2005; Tversky and Hard 2009). My theoretical development is also set apart from the ‘direct matching hypothesis’ (Wohlschlager et al. 2003) that is proposed as an explanation for the activations of motor repertoire triggered by the observation of actions and is proposed to be subserved by the mirror neuron system (Keysers and Perrett 2004; Rizzolatti and Craighero 2004). This entails an important issue as many treat those processes as embodied which involve any sort of bodily representations (whether afferent or efferent)2 but I side with the position adopted by Arzy et al. (2006). They define embodiment as involving the perception of one’s sense of self within the physical boundary of one’s body.

Future Directions

Separating egocentric reference frame from object based frames of reference should be an important pursuit for future scientific attempts. This is because people do not divide up the space that surrounds themselves in equal proportions. The status of the surrounding space is critically dependent upon the usual interactions of people within it. For instance, Franklin et al. (1995) showed that the region in front of our bodies is more important and could be described with the greatest detail. Thus, the conceptualization of our surrounding space is a function of our bodily asymmetries and the body’s relation with the world, forming the bedrock of our embodied abilities. Supporting results show that observers in the upright position are faster to identify objects from their vertical axis (head/feet), followed by the front-back horizontal axis while giving worst performance on the left-right horizontal axis (Franklin and Tversky 1990). This needs to be contrasted by Cartesian-like conceptualization of space (four regions of equal size) for neutral circular arrays (Huttenlocher et al. 1991).

Concluding Remarks

The current paper attempts to classify egocentric representations differently from the Traditional and Levinson’s terminology. The different classification is supported by the sub classification of egocentric representation into two distinct subtypes where the self/body plays an essential role. The first order representations are simple in nature yielding spatial localization with respect to one’s body while the second order representations are dependent upon the differences in viewing position(s) aiding egocentric transformation not only in the physical but also in the mental domain.


For the present concerns, such representations will be treated separately from any reference to the human body including the viewer’s perspective.


Personal communication by Kessler .



This work was supported by the Deutsche Forschungsgemeinschaft (DFG) grant managed through the Graduate School of the Centre of Excellence “Cognitive Interaction Technology”, University of Bielefeld, Germany.

Copyright information

© Springer Science+Business Media B.V. 2010