Construyendo mentes desarrollo de la comprensión de estados mentales en la ontogénesis típica y en niños con síndrome de Williams

Résumé

When somebody commences an adventure, even if it is as small as defining a research proposal, every piece of advice is welcomed. We can start following the advice by Karmiloff- Smith and Inhelder in 1974: ¿If you want to go ahead, get a theory¿. Theories enable us to demarcate a domain of interest and the way to approach it. As in Piaget and Inhelder¿s task of the three mountains (1948/1976), where the child has to take into account from where one is looking to know what is seen, the perspective taken will establish which portion of reality will be tackled and how it will be observed. If we want to go ahead we will have to determine a route. We are seeking a theory that enables us to analyse the path followed by the process of study (in this case mindreading abilities) as well as the trajectory of psychological functioning of the groups under study. In the first chapter we will attempt to define an ontogenetic perspective of study. This will be opposed to static cognitive neuropsychology, and in the description of each position some examples will be given on how they explain various developmental situations and which conclusions can be drawn from the two perspectives. Some of these examples will be referred to Williams syndrome (WS). In 1988 Bellugi and her team published a work on people with a WS cognitive profile. They found in this population a clear dissociation between intact linguistic abilities and severe cognitive deficits (Bellugi, Savo and Vaid, 1988). The message was clear: WS provides obvious proof on the independence of language and cognition. With this work Bellugi catapulted WS into the realm of interest of researchers concerned with the relationship (or lack of relationship) between language and cognition. These descriptions of so neatly defined dissociations set WS in a privileged position to be used as an argument to proposals suggesting the existence of cognitive modules with independent functioning, an innate origin, and that will appear selectively altered or preserved in this syndrome from birth. Descriptions of their strongest and weakest cognitive aspects from a relative point of view have been put forward as arguments favouring nativist approaches. Approaches that also establish a direct relationship between genotype and phenotype, so that the phenotypic result of a genetic disorder would be a reflection of its initial state. This analysis is based on a neuropsychological model of adult brain damage that has followed a typical development (TD). When a brain has developed in an habitual way and subsequently suffers a lesion, it is probable that some of its functions remain impaired and others intact. Adult cognitive neuropsychology is based on those dissociations to affirm that the brain is organised into cortical circuits of specific domains that work in a relatively modular mode. If the same logic is applied to the study of developmental disorders, in the event of similar dissociations, the atypical development is also explained in terms of impaired or intact modules, establishing its study as an argument against or in favour of affirmations of genetic pre-specification and modularity in the typical ontogeny. The continuity modular hypothesis would therefore disregard the role of development in attaining the phenotypic result, and would propose a brain organisation in innate mental (or neural) modules, present at birth with the same potentiality for dissociation through the lifespan, assuming a transparent relationship between the genotypic and phenotypic results. Applied to cognitive functioning in WS, this hypothesis would project a final state as a normal system with impaired and preserved components. Supporting their arguments in an interpretation of the data on WS in absolute instead of relative terms, this model will trace a cognitive profile in WS in which aspects such as visuospatial processing or numeric reasoning would be subtracted from the system. Meanwhile linguistic, face processing, and social comprehension abilities would appear selectively preserved. However, regarding WS or any other developmental disorder, some previous considerations should be borne in mind before making affirmations about the organisation of the developing cognitive system from adult neuropsychology descriptions. Models on adult functioning are not easily applicable to children, because the normative data on which they are based might not always be valid for a system in a constant¿but not homogeneous¿ process of change. However, paradoxically, research on typical and atypical cognition in children does not always follow an ontogenetic perspective (Karmiloff-Smith, 1997, 2000). Developmental cognitive neuropsychology (Temple, 1991) is developmental because it tackles the developing system, not the development of the system. In order to study the development of the system it is crucial to consider development as a cause, as means (Karmiloff-Smith, 1981), and as an aim in itself (Karmiloff-Smith, 1998), as the origin of the ultimate processes and also as a consequence of those processes. Piaget defined four factors responsible for mental development: biological factor, physical experience, social experience, and an inner construction mechanism. The former three take place in development; the fourth is the developmental process itself, which cannot be restricted to innateness or to a prespecified plan (Piaget and Inhelder, 1969). Karmiloff-Smith learnt from Piaget that ontogenetic development itself is the clue to understand the structure of the resulting cognitive system (Karmiloff-Smith, 1998a). From the neuroconstructivistic perspective (Karmiloff-Smith, 1998a; Mareschal, Johnson, et al. 2007) the brain does not start with pre-specified modules, aimed at the independent processing of specific cognitive domains, but nor is it a question of time for modularisation to make each brain area work in a specific way. The infant brain will follow a process of specialisation and localisation until the function is gradually established. Development is sustained in some restrictions as starting points, initially domain relevant, that would only become domain specific at the end of this developmental process, through specific interactions with the environment (Karmiloff-Smith, 1992a). A proposal of integration: processes of explicitation and specialisation as channels for development. Karmiloff-Smith (1992a) defined development on the basis of two parallel processes: a progressive modularisation and also a gradual explicitation. The neuroconstructivist perspective is aimed at explaining the process of emergent modularisation in detail. Through the progressive specialisation of representations, the functioning of the domain becomes faster and more efficient. Access to the information in the representations that enables behavioural success is reduced, so that representations in the modules would remain implicit for the system. Modular structures are formed from minimal architectural and computational constraints in interaction with the environment. In parallel, according to Karmiloff-Smith (1992a, 1998b, 2005), the process of representational change, by which the individual re-represents the knowledge already obtained, would enable greater access to the component parts of this knowledge, so that the system would become slower but more flexible. The gradual explicitation of representations would imply a continuous process, which can be organised into a hierarchy of mechanisms of increasing complexity (Pozo, 2003), culminating in the representational redescription process (RR). The RR is a general domain operation that acts in specific domain representations. The process by which ¿information in the mind subsequently becomes explicit knowledge to the mind¿ (Karmiloff-Smith, 1992a). Karmiloff-Smith¿s strategy was to go beyond behavioural mastery, studying those developmental patterns that describe an appropriate performance in the task on an initial level, a second moment in which the number of error increases, to get a third time when performance is correct again. Even if the first moment is similar to the last one on a behavioural level, for the RR model they are different on a representational level. The initial success would be obtained by means of a very overall representation that, once behavioural mastery is achieved, will start to specify, analysing the procedures in the performance and dividing them into components. The RR model first aims to offer a more dynamic conception development than Fodor¿s rigid modular proposal, and secondly, it suggests a model of phases that, unlike the Piagetian stages that depend on age and imply changes globally affecting the cognitive system, enables linking acquisition times of cognitive processes to the specific mechanism implied. To Karmiloff-Smith the development of these two processes makes human intelligence specifically human (1998b). This affirmation is one of the main foundations of this work: processes of modularisation and explicitation govern development. However, in our opinion, it also raises two important doubts: first, whether parallel development necessarily implies independence between the two processes. They co-occur at the same time but, in spite of the relationship between them not being evident (they could even be understood as opposite), it is necessary to analyse the possibility of their interdependence. Related to this, the second question would be whether all cognitive developments are based equally on each of the two processes. In order to continue thinking about these questions it could be helpful to tackle the taxonomy of psychological functions proposed by Rivière (1999/2003a). He describes their genesis and organises them as type 1 functions, (or modular non-specific), type 2 functions (or constructive non-social non-specific), type 3 functions (rudimentary-higher) and type 4 functions (or proper-high-functions). The four hierarchical and cumulative psychological functions can be described according to fifteen dimensions (Rivière, 1999/2003a). Regarding the description of the four functions on these dimensions, we believe they could be organised in relation to the continuum of dependence from the two processes proposed by Karmiloff-Smith. If we imagine two orthogonal vectors, so that one of them represents the emergent specialisation progress and the other the process of gradual explicitation, the different types of functions could be distributed on both vectors. Most of the properties that Rivière attributes to functions can be applied to the two processes (and in many cases they can be understood as the opposite extremes of both vectors). However, it is possible that some of those properties are better understood related to one of the processes. The features that arise from each type of function that might be related to the specialisation process would be: the degree of modularisation, their degree of genetic conditioning, ontogenetic and philogenetic development, dependence on critical periods, neural localisation, and their efficiency and flexibility. Functions would be distributed into a continuum in every of these characteristics. At one of the extremes, type 1 functions would be highly modular, would appear described and prescribed in the genome, would not be susceptible to interaction, their acquisition would imply a critical period, they would be associated with a precise neural localisation, and they would show inflexible and very efficient functioning. At the other extreme, development of type 4 functions would depend completely on culture and require explicit instruction; they would not imply neural pruning processes and therefore they would not have a precise neural localisation. According to a definition of efficacy as a quotient between cost and benefit, they would be much less efficient and more flexible, and there would not be critical periods in their acquisition. Although we are setting them on a continuum it is important to clarify that the four types of functions are not distributed in a uniform way on each of the vectors. Type 2 functions would be closer to type 1 than to type 3 functions in many of the dimensions of this specialisation vector (however, possibly it would not be the same for the dimension of modularity, which Rivière does not define expressly for type 3 functions). While type 4 functions would be allowed by the genome, type 3 functions would be defined by genes and formatted by culture. Their phylogenetic past would be much more extended in type 1 functions, while type 4 functions would be based on a historical origin. Only type 3 and 4 functions would be exclusive for humans. Although this division is completely artificial and only satisfies a stated proposal, defining characteristics for the functions that would be more closely related to the explicitation vector would be the ones describing its process of learning, their dependence on social relationship and culture, the demands of attention and conscious processing implied, and their symbolic character. Type 1 functions are not learned. In a strict sense, probably type 2 and type 3 functions are not learned either, but Rivière uses the term learning without its meaning of product of instruction. Type 2 functions follow a procedural learning process, type 3 functions follow a procedural-declarative learning process, type 4 functions are learned by declarative processes. Representations of type 3 functions are redescribed. Rivière states that they imply an implicit learning that, to a large extent, can be explained from Karmiloff- Smith¿s model (Rivière, 1999/2003b, p. 283). Representations of type 3 functions are explicitated gradually. The dependence on consciousness of type 4 functions is very high. However, not even in type 4 functions would it be possible to explicitate all the components in the process, and this will permit some efficacy. Their attention-related requirements would be also greater in higher functions, as well as their dependence on culture. Type 1 functions do not depend on culture. Not until type 3 functions can we speak of symbolic processing. Type 3 and type 4 functions imply a capacity of symbolisation, but the former are the first to establish symbols. Type 3 and 4 functions depend on interaction, but the shape this interaction has to adopt would be different: type 3 functions are acquired only in an upbringing context, while type 4 functions are learned in specific educational contexts. Rivière also describes the vulnerability of the four types of functions. Johnson, Karmiloff-Smith, Pennington & Oliver (2000) maintain that the neuroconstructivistic approach still has to answer some questions, and to explain for instance why some patterns of deficits are more common, why mental delay is a common result for so many etiologies, or why there is a lower incidence of disorders whose main deficits are at lower levels of processing. Type 1 and type 2 functions have a very low susceptibly to alteration; some developmental disorders are characterised by some specific deficits on type 3 functions, while difficulties in the acquisition of type 4 functions are broadly extended. Disorders whose core deficits are set at the lowest levels of sensory processing would correspond to deficits in type 1 and type 2 functions. Mental delay could be the consequence of a deficit in any of the functions. To summarise the representation of the two orthogonal vectors to the development of specialisation and explicitation, it would be possible that unspecific modular functions¿ closer to biology and with a more efficient, rapid and unconscious functioning¿could lean on the former vector; while proper high functions (on which Rivière maintains that there is no point in talking about modules) would lean fundamentally on vector RR. Type 3 functions would be localised precisely on this vertex between biology and culture, and they would be a product of the two vectors. These rudimentary high functions are those strictly defining human development (Rivière) that lean on these two processes (Karmiloff-Smith). Rivière¿s model of psychological functions implies an ontogenetic development, so that type 4 functions presuppose the development of the other three functions, and so on recursively. This could seem to suggest that in development the specialisation process is prior to the explicitation one. However, a better way to understand the trajectory of these two processes would be in mutual interdependence. Although in the global developmental course initial functions fundamentally follow a process of specialisation, and the development of the latter tends towards a gradual explicitation, regarding the development of each domain it would be necessary to understand the implication of both processes in a joint and interdependent way. In each particular domain the resultant representations would be the explicit redescription of the implicit knowledge in the developing modules. But in the dynamic of development, the progressive specialisation also requires the products of these redescriptions. Explicitation takes place on new implicit knowledge of a gradually more specific domain. If typical development runs on these two processes, then it would be probable that atypical development could be explained from the difficulties in any or both of them. Specifically, Karmiloff-Smith (1998b) proposes that in some cases (as in Down syndrome) cognitive problems could be caused by a failure in the modularisation process. This deficit in the specialisation process could be implying that the process of neural pruning was not taking place. On a cognitive level, a lack of modularisation in developmental disorders would indicate common processes among initially different domains, that would ultimately be modularised and divided. In other cases, however, specialisation could be occurring too rapidly, limiting the influence of environmental restrictions in the specification of the result. In atypical development, the repeated processing of determined inputs does not mean domain relevance is turned into domain specificity. This repeated processing of specific inputs is the requirement established by Karmiloff-Smith for the RR (1992b). Behavioural mastery is not a prerequisite for redescription, it would be sufficient with a situation of a stable state (even if repeated answers were erroneous). It would be sufficient that the system starts to capture regularities in the inputs and processes them repeatedly. Without this repeated processing, with a certain organisation function, of imposition of structure to the perception of these regularities (Pozo, 2003), there could be no subsequent redescription or specialisation of the system. Some of the difficulties the atypically developing system has to induce specialisation or explicitation processes could be set in a deficit in the processing of specific inputs. Developmental disorders could present a deficit either in the representational precursors, in the acquisition of subsequent knowledge, in the gradual specialisation process, or in both processes. It could be expected that the type of mental functions involved would depend on the impaired process and its degree of impairment, the atypical precursors, the implied process of knowledge acquisition, or the problem presented by a domain for its specialisation. From this speculation, and from the consistent arguments of the neuroconstructivistic perspective, we will try to characterise one of the psychological functions developed in the intersection between biology and culture, that are specialised and explicitated. We will try to explain mentalising abilities from its description as a type 3 function. Among the countless works on typical development of theory of mind, this will try to look at this competence from the study of its relationship with the processes of explicitation and specialisation.