Here we consider two apparently distinct questions: [1] What are the neural correlates of causal inference? And [2] How do we distinguish between different domains of causal inferences? To understand the varieties of causal thinking in human minds, we need to bring together behavioral and developmental data on the one hand and information from both neuro-psychology and neuro-imaging on the other. Once this evidence is replaced in an evolutionary framework, it becomes easier to understand the functional divisions between neural systems. We discuss these questions in the context, first, of high-level conceptual differences between living-things and artifacts, and then of low-level causal perception.

Computation, neural systems and evolution

In some domains, the mapping between computational function and neural structure seems rather straightforward. Consider face-recognition. There is substantial behavioral evidence to support the notion of a specialized computational device (Young et al., 1987; Tanaka & Sengco, 1997). Some functional features (e.g. configural processing, sensitivity to inversion) are found only in the treatment of face-like stimuli (see (Kanwisher, 2000)for a detailed discussion). There is also developmental evidence in infancy (Pascalis et al., 1995; Slater & Quinn, 2001). Neuro-psychology has documented many cases of prosopagnosia (Farah, 1994)in which the structural processing of objects, object-recognition and even imagination for faces can be preserved while face-recognition is impaired (Duchaine, 2000; Michelon & Biederman, 2003).

In this case, functional specialization happens to map onto to neural structure. Neuro-imaging studies have reliably shown a specific modulation of activity of the fusiform gyrus of the temporal lobe during identification or passive viewing of faces (Kanwisher et al., 1997). Specialized systems handle the invariant properties of faces (that allow recognition) while other networks handle changing aspects such as gaze, smile and emotional expression (Haxby et al., 2002).

The example of face-recognition is exceptional, in the sense that there is a clear mapping here between a unique and clearly defined function on  the one hand and a unique structure on the other. But the example is relevant here in that it shows how evolutionary considerations are the necessary background to functional distinctions. Let us consider this in more detail.

Despite the impressive behavioral and neural evidence, some psychologists have argued that the specificity of face-perception was an illusion, and that human beings simply became expert recognizers of faces by using unspecialized visual capacities. One central argument is that one can observe inversion effects (Diamond & Carey, 1986; Gauthier et al., 1998)and fusiform gyrus activation (Gauthier et al., 1999; Tarr & Gauthier, 2000)when trained experts examine and identify automobiles, birds, dogs or even abstract geometrical shapes (the effects occur only in their particular domain of expertise). So, the argument goes (we simplify a bit), the system is not actually dedicated to faces, but to a broader domain of (say) ‘visual stimuli of personal importance with similar overall structure and fine-grained distinctive features’.

There are three possible objections to this alternative interpretation, two of which are tempting but misguided:

[1] One may object that ‘face’ is a simpler concept that ‘visual stimuli of … etc.’ so we should prefer the first predicate, even if the cognitive system considered is often rather lax in its interpretation of what a ‘face’ is. However, this is misguided because there really is nothing intrinsically more parsimonious in one concept than the other. The complexity of predicates depends on a matrix of other predicates they are related to, so that ‘visual stimuli of … etc.’ may be the simpler predicate in some alternative ontology (Goodman, 1955).

[2] One may think that ‘face’ is a more natural predicate, because there are such things as faces in the world (people have faces, animals have faces), whereas the domain of ‘visual stimuli of… etc.’ does not to correspond to a proper natural kind. This too is misguided, because as it is difficult to argue that faces really are a natural kind of objects. Perhaps faces as such do not enter into any causal laws except in the computational states of face-recognizers… which leads us to the third argument:

[3] Being good at (or slightly better than others at) distinguishing conspecifics can have important fitness effects. Indeed, various human behaviors of great evolutionary significance (social exchange, friendship, cooperation, warfare, coalitional affiliation) depend on the precise identification of conspecifics. To the extent that the top front part of conspecifics (what we usually call “the face”) is used as a source of information for that purpose, ‘faces’ are part of the cognitive environment of human beings in a way that ‘visual stimuli of … etc.’ are not. Some species of primates evolved to become better recognizers of ‘faces’, not better recognizers of ‘visual stimuli of … etc.’ because the only domain where their performance mattered to fitness was the distinction between conspecifics.

This may seem rather obvious – because the functional specialization of face-recognition is not in fact really controversial. We emphasize the point, because such evolutionary considerations become even more important as we turn to domains where functional distinctions are less obvious.

Domains: Living things / artifacts-materials

Let us turn to the high-level distinction between the domains of ‘living things’ and ‘artifacts’, generally interpreted as a functional distinction based on specific causal principles. Animal species are intuitively construed in terms of species-specific “causal essences” (Atran, 1998). By contrast, man-made objects are principally construed in terms of their functions. (Kemler Nelson, 1995)(Richards et al., 1989). Artifacts seem to be construed by adults in terms of their designers’ intentions as well as actual use (Bloom, 1996)and pre-schoolers too consider intentions as relevant to an artifact’s ‘genuine’ function (Gelman & Bloom, 2000).

These are differences of inferential principles. The fact that an object is identified as either living or man-made leads to [a] paying attention to different aspects of the object; [b] producing different inferences from similar input; [c] producing categories with different internal structures (observable features index possession of an essence [animals] or presence of a human intention [artifacts]; [d] assembling the categories themselves in different ways (there is no hierarchical, nested taxonomy for artifacts, only juxtaposed kind-concepts).

The neuro-psychological evidence seems to supports this notion of distinct structures. Some types of brain damage result in impaired content or retrieval of linguistic and conceptual information in either one of the two domains. The first cases to appear in the clinical literature showed selective impairment of the living thing domain, in particular knowledge for the names, shapes orassociative features of animals (Warrington & McCarthy, 1983; Sartori et al., 1993; Sheridan & Humphreys, 1993; Sartori et al., 1994; Moss & Tyler, 2000). But there is also evidence for double dissociation, for the symmetrical impairment in the artifact domain with preserved knowledge of living things (Warrington & McCarthy, 1987; Sacchett & Humphreys, 1992).

However, it is very difficult in this case to map the functional distinctions and their behavioral manifestations onto neural structures or systems. In particular, the neuro-imaging evidence is less than altogether compelling. A host of neuro-imaging studies, using both PET and fMRI scans, with either word- or image-recognition or generation, have shown significantly different cortical activations for living things and artifacts (Martin et al., 1994; Perani et al., 1995; Spitzer et al., 1995; Martin et al., 1996; Spitzer et al., 1998; Moore & Price, 1999; Gerlach et al., 2000). However, few of these findings are clearly replicated. Also, in many studies the variety of activation peaks reported for either type of stimuli could not plausibly be described as constituting a functional network. That is, there is no clear indication that joint involvement of such areas is required for the processing of such stimuli. The gross anatomy does not suggest particular and exclusive connectivity between those regions either. Finally, some of the findings may turn out to be false positives (Devlin et al., 2002).

Reinterpretation from an evolutionary viewpoint

The problem in this case is that the domains themselves are not construed in a principled way. In most studies of domain-specificity, the precise understanding of what are ‘artifacts’ (often oddly called ‘objects’) or ‘animals’ or ‘living things’ is left to… the experimenter’s commonsense, as if that was a privileged road to cognitive structure. In some other cases experimenters stick to scientific distinctions between the ‘living’ and ‘non-living’, somehow implying that the organization of mental faculties corresponds to the way the world really is.

From an evolutionary viewpoint, we should expect cognitive domains to correspond to recurrent fitness-related situations or problems (e.g. ‘predators’, ‘competitors’, ‘tools’, ‘foraging techniques’, ‘mate selection’, ‘social exchange’, ‘interactions with kin’, etc.). This suggests that we should find systems specialized, not so much in different kinds of objects, but in different kinds of interaction with objects, likely to impinge on the organism’s fitness.

From that standpoint, humans certainly do not interact with “living things” in general. Living things comprise plants, bacteria, and middle-sized animals including human beings. We interact in very different ways with predators, prey, potential foodstuffs, competitors, and parasites. Nor do humans handle “artifacts” in general. Man-made objects include foodstuffs, tools and weapons, buildings and shelters, visual representations, as well as paths, dams and other modifications of the natural environment. We should expect the input format and activation cues of domain-specific inference systems to reflect this fine-grained specificity.

This hypothesis of a set of finer-grained, fitness-relevant systems receives some support from the neuro-imaging evidence. For instance, the naming of artifacts, or even simple viewing of pictures of artifacts, seems to result in pre-motor activation. Viewing an artifact-like object automatically triggers the search for (and simulation of) motor plans that involve the object in question. Indeed, the areas activated (pre-motor cortex, anterior cingulate, orbito-frontal) are all consistent with this interpretation of a motor plan that is both activated and inhibited. This suggests that “man-made object” is probably not the right criterion here. Houses are man-made but do not afford motor plans that include handling. If motor plans are triggered, they are about tools rather than man-made objects in general (Moore & Price, 1999). A direct confirmation can be found in a study of manipulable versus non-manipulable artifacts, which finds the classical left ventral frontal (pre-motor) activation only for the former kind of stimuli (Mecklinger et al., 2002). By contrast, some infero-temporal areas (BA20) are found to be exclusively activated by animal pictures (Perani et al., 1995), as are some occipital areas (left medial occipital) (Martin et al., 1996). The latter activation would only suggest higher modulation of early visual processing for animals. This is consistent with the notion that identification of different animal species requires finer-grained distinctions than that of artifacts: animals of different species (cat, dog) often share a basic Bauplan (trunk, legs, head, fur) and differ in details (shape of head, limbs, etc.), while tools (e.g. screwdriver, hammer) differ in overall structure. Animal-specific activations of the posterior temporal lobe seem to vanish when the stimuli are easier to identify (Moore & Price, 1999)which would confirm this interpretation as an effect of fine-grained, relatively effortful processing[i].

Neuro-imaging findings and developmental evidence converge in supporting the evolutionarily plausible view, that inference systems are not about ontological categories like “man-made object” or “living thing” but about types of situations, such as “fast identification of potential predator-prey” or “detection of possible use of tool or weapon”.

Neural evidence for causal perception

Let us now turn to causal perception, the fast and automatic interpretation of distinct visual events as causally related. The cognitive approach to causal perception stems from Hume – and from Michotte’s study of causal inferences for the trajectories of billiard-ball-like shapes (Michotte, 1946). The interpretation of such visual events as causal or non-causal depends on a subtle psycho-physics that combines time- or space-contiguity between events with relative velocities (Schlottman & Shanks, 1992)from infancy (Leslie, 1984; Schlottmann & Surian, 1999). To the extent that there is principled interpretation of particular stimuli, what specific neural structures are involved?

Surprisingly, this was not the object of much research until recently. One major reason may be that there seem to be no cases of selective impairment of causal reasoning. So there is no neuro-psychological evidence and no tentative connections to localized infarction or other damage. This may suggest that no specific structure is involved in causal perception, that the process is distributed among many different systems. Or it may suggest that the process in question is so fundamental to other (higher-level) domains of cognition that an impairment in this capacity would result in general confusion and therefore not to a diagnosis of selective impairment.

Be is as it may, there is very little neuro-imaging investigation of the networks modulated by causal perception. One exception is an event-related fMRI study by Blakemore and colleagues, with stimuli including various psycho-physical variants of causal and non-causal collisions (Blakemore et al., 2001). Results of the causal – noncausal subtraction show higher activation MT/V5, STS (superior temporal sulcus) and the left IPS (intra-parietal sulcus). Finding activation in MT/V5 is not in itself surprising (the region is sensitive to relative motion) but the difference in activation may suggest that a causal event is more complex or triggers more processing than a non-causal event. The other two regions are generally involved in the higher-level interpretation of visual events. Taken together, these activations suggest that causal events are treated as special from the lowest stages of visual processing. But note that the system involved is not a “Hume module” for causal inference in general but rather a detector of specific visual psycho-physics.

Pure causal processing or joint activation?

The search for a “Hume module” may be a wild goose chase because the system’s putative domain of operation is unconstrained. Do organisms really need to process causal events as such and distinguish them as a general class from non-causal events? This would be self-evident if brains had been designed to be philosophically correct, as it were. But they were designed to enhance fitness.

From this standpoint, the mere detection of contingencies in one modality may be of great interest to the extent that it can feed into specialized downstream processing. That is to say, the quick perception of two events as contiguous (which may or may not involve cortical structures) would be quickly followed by activation of possibly relevant systems specialized in fitness-relevant situations. For instance, it seems plausible that the detection of potential predators occurs as a fast response to specific kinds of unexpected stimuli (e.g. sudden noise without visual correlate). Another kind of detection is involved in the way organisms monitor the effect of their own actions on objects. Specific parameters of the contingency relations between events may trigger activation of such a detection system.

All this is necessarily rather speculative, but would at least suggest that it is not the contingency detectors themselves that produce causal inferences, but the joint activation of contingency-detection and situation-relevant inference engines. This would have interesting consequences. For instance, it would suggest that causal inferencing in the brain rarely is a matter of single-modality information-processing, but usually involves comparison of several modalities. Although there is evidence that cross-modality contingency detection is a fast and automatic system (Driver & Spence, 1998; Schmitt et al., 2000), there is no study of such effects in specific, biologically relevant situations.

Let us mention a last example that supports this notion of a joint activation of contingency detection and higher-level templates for causal inference. Blakemore and colleagues compared the neural activities triggered by both mechanical and “intentional” causal events between shapes on a screen (Blakemore et al., 2003). In the intentional condition one shape oriented in such a way as to “attend” to another shape, a much simpler intentional event than the familiar Heider films of triangles and squares chasing and helping each other (Heider & Simmel, 1944). Subjects either attended to the direction of motion or to the contingency relations. Results showed some interesting differences between mechanical and intentional events. But more interesting for our discussion here was the difference within the intentional condition, depending on the subject’s focus of attention. When subjects focused their attention on the contingent relationships between the objects in the displays, as opposed to physical aspects of the objects’ movement, there was significant activation of the right middle frontal cortex and the left STS, a subset of the regions that are consistently activated by theory of mind tasks (Fletcher et al., 1995; Brunet et al., 2000; Castelli et al., 2000; Gallagher et al., 2000; Vogeley et al., 2001). Although this activation had been previously described as an automatic result of “intentional-looking” contingencies (Happé et al., 1999), it seems more accurate to describe it as the effect of contingencies in the context of intentional expectations (Blakemore et al., 2003).In other words, given a biologically relevant situation (looking for possible animate agents in an unknown environment), the detection of contingencies involves both lower-level and specific higher-level neural systems. Rather than a Hume system, this is a “look-for-animate-agents” system.

Conclusions

There are obvious limits to this picture, notably because neuro-psychological cases are ambiguous and neuro-imaging techniques are limited. But the evidence reviewed here should point to a general lesson.

Evolutionary considerations are not a complement or footnote to the study of cognitive function, a sort of “by the way, this is how the system got to be this way…” commentary on prior results, independently produced by neuro-psychology or neuro-science or experimental psychology. A description of the “proper domains” of function is intrinsically an evolutionary description (Sperber, 1996). In the near future, we may expect to understand much better how the brain handles information about causal relations. The price for this better picture may be to jettison some philosophical baggage, such as the notion that organisms detect “causation” in general. Contrary to this “philosophical” way of proceeding, we should start from the computational requirements of specific fitness-relevant situations, predict which kinds of contingencies are pertinent to each specific type, and test for the existence and autonomy of corresponding neural systems. The examples reviewed here suggest that this may be a more successful strategy in the description of causal thinking.

References

Atran, S. A. (1998). Folk biology and the anthropology of science: Cognitive universals and cultural particulars. Behavioral & Brain Sciences, 21(4), 547-609.

Blakemore, S.-J., Boyer, P., Pachot-Clouard, M., Meltzoff, A. N., & Decety, J. (2003). Detection of contingency and animacy in the human brain. Cerebral Cortex, 13, 837-844.

Blakemore, S.-J., Fonlupt, P., Pachot-Clouard, M., Darmon, C., Boyer, P., Meltzoff, A. N., et al. (2001). How the brain perceives causality: An event-related fmri study. Neuroreport: For Rapid Communication of Neuroscience Research, 12(17), 3741-3746.

Bloom, P. (1996). Intention, history and artifact concepts. Cognition, 60, 1-29.

Devlin, J. T., Russell, R. P., Davis, M. H., Price, C. J., & Moss. (2002). Is there an anatomical basis for category- specificity? Semantic memory studies in pet and fmri. Neuropsychologia, 40(1), 54-75.

Diamond, R., & Carey, S. (1986). Why faces are and are not special: An effect of expertise. Journal of experimental psychology General, 115(2), 107-117.

Driver, J., & Spence, C. (1998). Crossmodal attention. Current Opinion in Neurobiology, 8(2), 245-253.

Duchaine, B. C. (2000). Developmental prosopagnosia with normal configural processing. Neuroreport: For Rapid Communication of Neuroscience Research, 11(1), 79-83.

Farah, M. (1994). Specialization within visual object recognition: Clues from prosopagnosia and alexia. In G. R. Martha J. Farah (Ed.), The neuropsychology of high-level vision: Collected tutorial essays. Carnegie mellon symposia on cognition. (pp. 133-146): Lawrence Erlbaum Associates, Inc, Hillsdale, NJ, US.

Gauthier, I., Tarr, M. J., Anderson, A. W., Skudlarski, P., & Gore, J. C. (1999). Activation of the middle fusiform "face area" increases with expertise in recognizing novel objects. Nature Neuroscience, 2(6), 568-573.

Gauthier, I., Williams, P., Tarr, M. J., & Tanaka, J. (1998). Training "greeble" experts: A framework for studying expert object recognition processes. Vision Research Special Issue: Models of recognition, 38(15-16), 2401-2428.

Gelman, S. A., & Bloom, P. (2000). Young children are sensitive to how an object was created when deciding what to name it. Cognition, 76(2), 91-103.

Gerlach, C., Law, I., Gade, A., & Paulson, O. B. (2000). Categorization and category effects in normal object recognition: A pet study. Neuropsychologia, 38(13), 1693-1703.

Goodman, N. (1955). Fact, fiction and forecast. Cambridge: Harvard University Press.

Haxby, J. V., Hoffman, E. A., & Gobbini, M. I. (2002). Human neural systems for face recognition and social communication. Biological psychiatry, 51(1), 59-67.

Heider, F., & Simmel, M. (1944). An experimental study of apparent behaviour. The American Journal of Psychology, 57, 243-259.

Kanwisher, N. (2000). Domain specificity in face perception. Nature Neuroscience, 3(8), 759-763.

Kanwisher, N., McDermott, J., & Chun, M. M. (1997). The fusiform face area: A module in human extrastriate cortex specialized for face perception. Journal of Neuroscience, 17(11), 4302-4311.

Kemler Nelson, D. G. (1995). Principle-based inferences in young children's categorization. Revisiting the impact of function on the naming of artefacts. Cognitive Development, 10, 347-380.

Leslie, A. M. (1984). Spatiotemporal continuity and the perception of causality in infants. Perception, 13(3), 287-305.

Martin, A., Haxby, J. V., Lalonde, F. J., Wiggs, C. L., Parasuraman, R., & Ungerleider, L. G. (1994). A distributed cortical network for object knowledge. Society for Neuroscience Abstracts, 20(1-2), 5.

Martin, A., Wiggs, C. L., Ungerleider, L. G., & Haxby, J. V. (1996). Neural correlates of category-specific knowledge. Nature (London), 379(6566), 649-652.

Mecklinger, A., Gruenewald, C., Besson, M., Magnie, M.-N., & Von Cramon, Y. (2002). Separable neuronal circuitries for manipulable and non-manipulable objects in working memory. Cerebral cortex, 12, 1115-1123.

Michelon, P., & Biederman, I. (2003). Less impairment in face imagery than face perception in early prosopagnosia. Neuropsychologia, 41(4), 421-441.

Moore, C. J., & Price, C. J. (1999). A functional neuroimaging study of the variables that generate category-specific object processing differences. Brain, 122(5), 943-962.

Moss, H. E., & Tyler, L. K. (2000). A progressive category-specific semantic deficit for non-living things. Neuropsychologia, 38(1), 60-82.

Pascalis, O., de Schonen, S., Morton, J., Deruelle, C., & et al. (1995). Mother's face recognition by neonates: A replication and an extension. Infant Behavior & Development, 18(1), 79-85.

Perani, D., Cappa, S. F., Bettinardi, V., Bressi, S., Gorno-Tempini, M., Matarrese, M., et al. (1995). Different neural systems for the recognition of animals and man-made tools. Society for Neuroscience Abstracts, 21(1-3), 1498.

Richards, D. D., Goldfarb, J., Richards, A. L., & Hassen, P. (1989). The role of the functionality rule in the categorization of well-defined concepts. Journal of Experimental Child Psychology, 47, 97-115.

Sacchett, C., & Humphreys, G. W. (1992). Calling a squirrel a squirrel but a canoe a wigwam: A category specific deficit for artefactual objects and body parts. Cognitive Neuropsychology, 9, 73-86.

Sartori, G., Coltheart, M., Miozzo, M., & Job, R. (1994). Category specificity and informational specificity in neuropsychological impairment of semantic memory. In C. Umilta & M. Moscovitch (Eds.), Attention and performance 15: Conscious and nonconscious information processing. (pp. 537-550). Cambridge, MA, US: The MIT Press.

Sartori, G., Job, R., Miozzo, M., Zago, S., & et al. (1993). Category-specific form-knowledge deficit in a patient with herpes simplex virus encephalitis. Journal of Clinical & Experimental Neuropsychology, 15(2), 280-299.

Schlottman, A., & Shanks, D. R. (1992). Evidence for a distinction between judged and perceived causality. Quarterly Journal of Experimental Psychology: Human Experimental Psychology, 44(2), 321-342.

Schlottmann, A., & Surian, L. (1999). Do 9-month-olds perceive causation-at-a-distance? Perception, 28(9), 1105-1113.

Schmitt, M., Postma, A., & De Haan, E. (2000). Interactions between exogenous auditory and visual spatial attention. Quarterly Journal of Experimental Psychology: Human Experimental Psychology, 53A(1), 105-130.

Sheridan, J., & Humphreys, G. W. (1993). A verbal-semantic category-specific recognition impairment. Cognitive Neuropsychology, 10(2), 143-184.

Slater, A., & Quinn, P. C. (2001). Face recognition in the newborn infant. Infant & Child Development Special Issue: Face Processing in Infancy and Early Childhood, 10(1-2), 21-24.

Sperber, D. (1996). Explaining culture: A naturalistic approach. Oxford: Blackwell.

Spitzer, M., Kischka, U., Gueckel, F., Bellemann, M. E., Kammer, T., Seyyedi, S., et al. (1998). Functional magnetic resonance imaging of category-specific cortical activation: Evidence for semantic maps. Cognitive Brain Research, 6(4), 309-319.

Spitzer, M., Kwong, K. K., Kennedy, W., & Rosen, B. R. (1995). Category-specific brain activation in fmri during picture naming. Neuroreport: An International Journal for the Rapid Communication of Research in Neuroscience, 6(16), 2109-2112.

Tanaka, J. W., & Sengco, J. A. (1997). Features and their configuration in face recognition. Memory & Cognition, 25(5), 583-592.

Tarr, M. J., & Gauthier, I. (2000). Ffa: A flexible fusiform area for subordinate-level visual processing automatized by expertise. Nature Neuroscience, 3(8), 764-769.

Warrington, E. K., & McCarthy, R. (1983). Category-specific access dysphasia. Brain, 106, 859-878.

Warrington, E. K., & McCarthy, R. (1987). Categories of knowledge: Further fractionations and an attempted integration. Brain, 110, 1273-1296.

Young, A. W., Hellawell, D., & Hay, D. C. (1987). Configurational information in face perception. Perception, 16(6), 747-759.