COGNITIVISMO, MENTE Y COGNICIÓN (Algunos conceptos y definiciones)

En principio, el término cognición en el contexto de las ciencias cognitivas hace referencia a una serie de capacidades como percepción, memoria, atención, lenguaje, aprendizaje y resolución de problemas. Por otro lado, la perspectiva cognitivista común se caracterizó por su asunción de una teoria computacional de la mente, cuyos elementos fundamentales serían dos:
- Reglas o algoritmos.

- Símbolos o representaciones simbólicas.

En su manual de 1967 (Cognitive Psichology) Neisser señalaba:  "Tal como se emplea aquí, el término “cognición” se refiere a todos los procesos mediante los cuales el ingreso [input] sensorial es transformado, reducido, elaborado, almacenado, recobrado o utilizado" (Psicología Cognoscitiva, p. 14). 

   Según Lycan (en Guttenplan, p. 319) el cognitivismo implica básicamente que:
- La explicación psicológica remite a estados y episodios internos.
- Los seres humanos y otros organismo psicológicos deben ser entendidos como sistemas de procesamiento de información. (Pero ¿qué es información?. Una aproximación interesante al concepto de información es la de Dretske en Knowledge and the Flow of Information).



En neurociencia cognitiva, lo "cognitivo" hace referencia, básicamente, a la dimensión psicológica o funcional (frente a la estructural y neurobiológica).


En cuanto a la diferencia entre mente y cognición, no siempre es clara o necesaria, pero si se busca una distinción mi propuesta provisional sería limitar la cognición a determinados procesos inconscientes (generalmente computacionales). En este sentido, la mente supondría añadir la conciencia y las emociones -aunque la separación entre cognición y emoción sea, de hecho, problemática-.

El concepto de cognición también es relevante en el debate sobre conciencia de acceso/conciencia fenoménica (Block, Lamme, Dehaene, etc.). En este debate la cognición suele aparecer asociada a la noción de acceso. Los defensores del enfoque disociacionista (es decir, partidarios de  disociar el acceso de la fenomenalidad: Block, Lamme, etc.) separan la fenomenalidad ("qualia", en un sentido amplio, o lo que otros llaman experiencia subjetiva) de la cognición.


Thagard (2005)
Esta autor ha caracterizado claramente la cognición y la mente (no habría aquí distinción entre ellas) en términos de representación y computación. Ha propuesto la expresión CRUM (Computational-Representational Understanding of Mind) para denominar la hipótesis central de la ciencia cognitiva: la mejor forma de entender el pensamiento es en términos de estructuras representacionales en la mente y de procedimientos computacionales que operan sobre esas estructuras (p. 10).
[Cf. la noción de GOFAI (Good Old-Fashioned Artificial Intelligence) propuesta por Haugeland]

Clark (2001) (Midware)
Explicó así la concepción de la mente que surgió en los primeros tiempos de las ciencias cognitivas: "La mente no era una cosa fantasmagórica, sino la operación de un sistema computacional formal implementado en el meatware del cerebro [...] El cerebro puede ser la implementación (local, terrenal, biológica) estándard, pero la cognición es una realidad a nivel de programa" (p. 13).

Haugeland (1981) (En Having Thought)
"Cognitivism, then, can be summed up in a slogan: the mind is to be undertood as an information processing system" (p. 26).
"Cognitivism is heir to this tradition [racionalismo]: to be intelligent is to be able to manipulate (according to rational rules) "clear and distinct" quasilinguistic representations -only now they´re sullied by omissions, probabilities, and heuristics. Deported from the immortal soul, however, they forfeit their original epistemic anchorage in the honesty of God and the natural light of reason [...] (pp. 39-40).

Shapiro (2011)
Ha señalado el carácter internalista del cognitivismo clásico (como postura tradicional de las ciencias cognitivas): "Las tareas cognitivas tienen un punto de inicio y de final determinado. Puesto que la cognición empieza con un imput hacia el cerebro y termina con un output desde el cerebro, la ciencia cognitiva puede limitar sus investigaciones a procesos dentro de la cabeza, sin tener en cuenta el mundo más allá del organismo" (p. 27).

Menary (2006)
Cuando Menary explica su propuesta "integracionista" recurre frecuentemente a la noción de "tarea cognitiva". Por ejemplo: "Nos implicamos en prácticas cognitivas, manipulando vehículos externos, para completar tareas cognitivas y esta es la marca de lo cognitivo" (p. 330). En una nota aclara: "Una definición general de una tarea cognitiva puede facilmente resultar vacía y poco útil. Si definimos la tarea cognitiva como cualquier tarea para cuya realización es necesaria la cognición, entonces casi cualquier tarea será cognitiva. Creo que es más útil pensar en tareas cognitivas como aquellas que implican el ejercicio de una particular capacidad cognitiva, como recordar una fecha, resolver un problema, aprender a conducir, etc. Estas son tareas dónde el ejercicio de las capacidades cognitivas está directamente ligado a su realización satisfactoria (to their successful completion)" (p. 343). Vemos, por tanto, que la noción de tarea cognitiva remite a la de capacidad cognitiva, que tampoco es definida, sino ilustrada mediante ejemplos.


Adams y Aizawa (varios)
En el debate en torno a la mente extendida Adams y Aizawa han propuesto una idea sobre qué es cognitivo: "the mark of the cognitive" ("a rather orthodox theory of the nature of the cognition" -2001, p. 52). En síntesis, la idea es que lo cognitivo supone un tipo especial de procesos que implican "representaciones no-derivadas" (p. 53). Estas representaciones se caracterizan por poseer un contenido intrínseco [sobre la noción de contenido en filosofía de la mente v., la entrada correspondiente en este blog]. Las representaciones no derivadas dependen del cerebro, lo que lleva a los autores a rechazar las concepciones externalistas (Clark, etc.) y a defender el "intracranealismo cognitivo contingente".
-"A Žfirst essential condition on the cognitive is that cognitive states must involve
intrinsic, non-derived content" (Adams y Aizawa 2001, p. 48).
Hay una segunda condición que hace referencia al tipo de procesos, pero más que identificar o definir el tipo de procesos que son cognitivos, lo que los autores hacen es señalar las diferencias entre procesos internos y procesos extendidos (que involucran recursos externos). [Este criterio parece incurrir en petición de principio ya que lo que se afirma es que los procesos implicados en la actividad cognitiva intracraneal son diferentes de los procesos implicados en la actividad cognitiva que utiliza herramientas externas, por tanto -esta sería presumiblemente la inferencia-, no hay cognición transcraneal. Pero esto supone asumir que solo son cognitivos los procesos internos. La diferencia de procesos por sí misma no demuestra nada, a no ser que se justifique que esta diferencia determina la frontera entre lo que es cognitivo y lo que no lo es, pero únicamente se dice que algo no es congitivo porque involucra procesos diferentes a los internos].

Menary (2010) Introduction to the special issue on 4E cognition
Menary defiende una noción de cognitivismo y de cognición en contraste explícito con la concepción de Adams y Aizawa: entiende la cognición simplemente como manipulación de representaciones. Y puntualiza: "I do not think that cognitivism is committed to anything stronger than this. Indeed, it would be unwise for cognitivists to hold to Adams´ and Aizawa´s stronger version of cognitivism -that only manipulations of representations with underived content count as cognitive- as this leads to the counter-intuitive conclusion that the many example of empirical work on brain´s transformation by internalising public symbol systems do not count as cognition" (p. 459, n. 2). 
Menary también distingue entre cierto cognitivismo débil, según el cual la cognición implica a veces representaciones (Clark, Sutton), y posiciones enactivistas (Chemero, Hutton).

Sutton et al. (2010)
Los autores se ven a sí mismos como ocupando un terreno intermedio (que compartirían con Clark y otros) entre la versión internalista del cognitivismo y el externalismo anti-cognitivista (p. 529). Sutton et al., mantienen una versión de la teoría representacional-computacional de la mente. Esta es la razón -afirman- por la que son a veces criticados por ciertos anti-cognitivistas extremos, quienes les acusarían de no proponer que la misma idea de cognición es un error y de no renunciar a la ciencia cognitiva (el presunto extremista es Button 2008, y también se mencionan Malafouris 2004 y Dreyfus 2007) (p. 528).

Considerando la noción de cognitivismo que aparece en Adams y Aizawa como aquella postura según la cual la cognición supone ciertas formas de manipulación de representaciones no derivadas, los autores afirman que tal noción es aceptada por Clark, pero que éste añade que la cognición también supone frecuentemente manipulaciones de representaciones derivadas (p. 528, n. 12).

Chemero (2009)
"Radical embodied cognitive science, very roughly, is the thesis that cognition is to be described in terms of agent-environment dynamics, and not in terms of computation and representation" (p. x).

Estados cognitivos frente a estados perceptivos (etc.) (Bayne)



Cognición y Función



SOBRE LA DIFERENCIA ENTRE MENTE Y COGNICIÓN (lo mental y lo cognitivo)
Adams y Aizawa (2001)

"The traditional question in the philosophy of mind is the mark of the mental. Here
we assume without argument that the mental is not the same as the cognitive, hence
that the mark of the mental is not the same as the mark of the cognitive. We take
learning, remembering, sensing, perceiving, and thinking to be paradigm cases of
cognitive processing. Further, we assume that qualia and phenomenal consciousness,
while real and associated with cognitive process, are not in themselves essential
elements of cognitive processes. Thus, we assume without argument that there can
be cognitive processes that lack associated quale and are not phenomenally conscious
to any agent" (p. 48).
Cf., con la posición de Clark frente a Noë (en la versión de Clark es dudoso que la mente extendida abarque también la conciencia).

Hutto (2011)
En su reseña de The Extended Mind (Menary -ed.,- 2010), Hutto define el externalismo activo (Chalmers y Clark) mediante cierta relación (y diferencia, por tanto) entre lo mental y lo cognitivo (o entre mente y cognición): El externalismo activo es la concepción "que sostiene que los procesos cognitivos subyacentes que hacen posibles ciertas ocurrencias mentales no están, al menos en algunas ocasiones, enteramente confinadas al cerebro" (p. 785). Y algo más adelante: "Sostienen [los defensores de la hipótesis de  la mente extendida] que cuando se cumplen ciertos criterios específicos los procesos cognitivos activos en los que superviene la mente se extienden al entorno" (p. 786).
Por tanto, parece que se trata de dos niveles distintos siendo el nivel "bajo" o subyacente el cognitivo y el "superior" el mental.

Clark y Chalmers (1998)
Titulan una de sus secciones "From Cognition to Mind". Es cuando introducen el ejemplo del cuaderno de Otto (es decir, "de Tetris a Otto").

Como yo lo veo
La distinción entre mente y sistema cognitivo (dejando aparte la cuestión de la conciencia) podría corresponderse con un nivel metafísico y un nivel (mientras se me ocurre un término mejor) científico-instrumental. Desde este punto de vista, un sistema cognitivo extendido no tendría por qué implicar necesariamente una mente extendida.

"True Believers: The Intentional Strategy". Daniel Dennett (en Lycan and Prinz 2008)

-Defensa (moderada -v. la distinción entre las dos afirmaciones empíricas-) del modelo computacional de la mente y del "language of thought".
-Me siguen pareciendo oscuras las nociones de "isomorfismo" y "reflejo" (mirror).
-Interesante revisar y analizar el concepto de representación.
-¿Qué aporta el ejemplo del termostato al debate?
-Acertadamente distingue entre dos afirmaciones empíricas:

“What Might Cognition Be, If Not Computation?” Tim Van Gelder

Probably the significance of this article is due to fact that it provides a potential answer to the “what else could it be” argument, since the goal of the author is to propose a viable alternative to the computational paradigm in cognitive science (346, 359).  According to this alternative, cognitive systems may be dynamical systems[AM1] . One of the most relevant features (if not the most) of this view is that, in contrast with orthodox-computational cognitive science, it eliminates representations -more precisely: at least certain kinds of systems are better understood excluding any reference to representations-.

The core of the argument deal with a device –called governor- of which purpose is to regulate steam power. This devise can be describe either in computational terms, that is, in form of an algorithm involving the manipulation of symbolic representations, or in a non-representational way, which coincides –this is a good rhetorical effect- with the actual solution to the problem, namely, the actual device built by James Watt.

The relation between certain parts of the device, the arm angle and the engine speed, is a key factor in the machine operation. According to the author this relation is a particular kind of dependence, which is non-representational. In fact, in order to describe such interaction a more powerful conceptual framework is needed, this framework is the mathematical language of dynamics.  The relevant suggestion whit regard to our course is that “Cognitive systems may in fact be dynamical systems, and cognition the behaviour of some (non-computational) dynamical system” (358). The crucial point is how to establish “the bridge”, that is, the relevant relationship, between the Watt device and the human cognition. This connection is provided by the motivational oscillatory theory (MOT) (361) [It is difficult for me to follow the mathematical explanation], mainly because “In MOT, cognition is not the manipulation of symbols, but rather state-space[A2]  evolution in a dynamical system” (362). (MOT is a case within a more general dynamical framework called for some authors “decision field theory”).

As far as I understand the argument the key feature is expressed in this sentence: “And it would be a model in which the agent and the choice environment, like the governor and the engine, are tightly interlocked” (360). 

CONSEQUENCES REGARDING EMBODIED COGNITION

We can call this point “The autonomy question”: To what extent is the brain autonomous or self-contained regarding cognition? This is a significant issue in the Gursh reply. Van Gelder[AM3] , indicates that given that according to the computational view “the cognitive system traffics only in symbolic representation, human body and the physical environment can be dropped from consideration; it is possible to study the cognitive system as an autonomous, bodiless, and worldless system whose function is to transform imput representation into output representations”  (373). Conversely, for the dynamical conception “the cognitive system is not just the encapsulated brain; rather, since the nervous system, body, and the environment are all constantly changing and simultaneously influencing each other, THE TRUE COGNITIVE SYSTEM IS A SINGLE UNIFIED SYSTEM EMBRACING ALL THREE” (373, see also 379).

(This is an expression of the continuous, simultaneous and mutually determining change which characterizes a dynamical system).

Another argument, among others, for the biological plausibility of this model concerns time. The role of time is more explicit and relevant in dynamics than in computational models, and “Timing is critical to a system that operates in a real body and environment” (379).

-Concerning the Cartesian assumptions considered by Grush, Van Gelder presents his notion of cognition as embodied and embedded in contrast to the Cartesian view and in concordance with the anti-Cartesian movement. In this sense, the dynamical account of cognition may be understood as a post-Cartesian approach to the mind and to the nature of the human being (see 380-381).

-Interesting remark: connectionist models are only a particular subcategory of dynamical systems. The kind of dynamical models suggested by the author for the cognitive system are different from connectionist model (see 370, 371)

SOME QUESTIONS

-In these two articles the notion of representation plays a significant role,  I find that a potentially problematic aspect of the nature of representation is the relation between representation and symbol. Van Gelder talks of “symbolic representation” (350, 353), in a sense in which is difficult to distinguish between symbol and representation. Is every representation symbolic?  On the other hand, the representational function appears to be a necessary element in the definition of symbol.

-When the author describes the computational model, he recalls that computation implies to manipulate symbols, and points out, that these symbols “have meaning” (350).  Does this mean that the machine (the device or its parts) work with semantic properties?

-According to Van Gelder if a system is not representational then it is not computational. Can be possible computation without representation? On the other hand, the author suggests the possibility that a dynamical system incorporates representations within a non-computational framework (376, 377).

-“The computer does not realize the abstract dynamical model; rather, it simulates it” (369). What is the difference between realization and simulation?

-Finally, What is the strongest argument to claim that human cognition is a dynamical system?

---

 [AM1]Dynamical systems are defined “as state-dependent systems whose states are numerical (in the abstract case, these will be numbers, vectors, etc.; in the concrete case, numerically mensurable quantities) and whose rule of evolution specifies sequences of such numerical states” (368).

 [A2]the notion of state-space: The concept of state-dependent system:  “A (concrete) state-dependent system Is a ser of features or aspects of the world which change over time interdependently…” (363).

 [AM3]Another question could be whether the modularity described in pp. 372-373 is correct.

 FURTHER QUESTIONS ABOUT DYNAMICAL SYSTEMS:

The mathematical nature of dynamical systems (“in the concrete case, numerically measurable quantities” -368-) arises some questions:

-It is possible to account for consciousness by means of a set of equations?

-Is this model compatible with the free will?

Can this model account for the “interface problem” (the relationship between personal and subpersonal levels?

-Does follow –obey- both the brain and the mind –that is, physical and mental properties- the same equations?

-If the operation of the mind is capable of being expressed in a set of equations does entail the existence (and the identification) of psychological laws?

-To what extent criticism over the crucial example of dynamical research on human decision making (basically MOT) can affect to the whole theory regarding cognitive science?

One important point in this sense could be: MOT is characterised in this way: “The framework thus includes variables for the current state of motivation, satiation, preference, and action (movement), and set of differential  equations describe how these variables change over time as a function of the current state of the system” (361).

 How the variables are identified and picked up (chosen)? And how the numerical value is assigned to every variable?  I think that this is a crucial problem since the model does not provide an answer and given that it is critical en the final result. Similarly, to contemplate an entire cognitive system (not just a particular task, skill or behaviour), for example the brain-body-environment system, or merely the mind-brain system, as a dynamical system arises the same questions.

---

 [AM1]Dynamical systems are defined “as state-dependent systems whose states are numerical (in the abstract case, these will be numbers, vectors, etc.; in the concrete case, numerically mensurable quantities) and whose rule of evolution specifies sequences of such numerical states” (368).

 [AM2]Another question could be whether the modularity described in pp. 372-373 is correct.

“In Defense of Some `Cartesian` Assumptions Concerning the Brain and Its Operation”. Rick Grush (2003)

DRAFT (to be proofread)

Important point: to look for an alternative model to the computational theory (in other “machines”)

The author argues against two main theses related to the embodied cognition conception (and to other close approaches in cognitive science), claimed by who Crush called “the radicals” (Haugeland, Brooks, Clark, Van Gelder and Thelen, among others):

-The mind is not in the brain: cognition involves the interaction between brain, body and environment in such a way that the brain cannot be separate from this system [I will dub it “thesis BBE”].

-Cognition does not require representation [I will call it “thesis NR”].

The `Cartesian` assumptions will be the two opposite ideas: the mind have and manipulate representations [perhaps the claim here is too strong], and the mind is autonomous. But at the same time the author rejects standard cognitive science, in particular symbolic computationalism, so REPRESENTATION WITHOUR SYMBOLS (or without symbolic computation). But, what is the meaning of `representation` in this context? The answer is provided by the emulation theory: roughly speaking, the brain have two parallel functions, one as a controller and other as a emulator, the representations are embodied in the emulator

“The world is not enough” (86) (Cf. Brook)

Key argument against the view of representation in Gelder: p. 65

Brook: mechanisms interacts directly with the world, they don’ t need a kind of “in-between” construction (that is, representations) (see 68).

I am specially concern in the role of neuroscience in the enquiry on mind, and a significant feature of this article is the attention drawn [participio pasado comprobar] to real biological systems, particularly to the brain, along with some suggestions content about neuroscience (see 64).

The emulation framework is a kind of information processing strategy that in fact is implement in the nervous system: “The main idea is that musculoskeletal emulator can process efferent copies and provide predictions of what the peripheral signal will be before the real peripheral signal is available” (76). Evidence Gush mentions to support his claim is that certain movement errors have been explained  (Wolpert, Ghahramani and Jordan, 1995)  “on the assumption that during the initial phases of a movement, before proprioceptive feedback is available, the motor centers exploit feedback from an internal model of the musculo-skeletal system, while as time progresses, feedback from the periphery is incorporated into the estimate” (77).  This internal model would be the emulator [related also to the simulation theory]

Another empirical evidence: Visual imagery (78) and visual perception (see Kosslyn)

PROBLEM: difference between on-line and off-line emulator operation  

PROBLEM: How the argument about emulator and representation match with the argument about components

QUESTION: the argument based on the plug criterion and the pilot example (80) is enough to conclude that “the brain is a notionally self-contained   locus of representational efficacy, not something that is essencially dependent upon the environment with which it is interacting…” (81). How the fact that the brain contains both the controller and the emulator is crucial for the brain to be a self-contained system?

The key point, I think, is that the emulator can operates without interaction with the real world

The argument rests on the emulation framework (74), in which is based the emulation theory of representation “according to which the brain constructs inner dynamical models, or emulators, of the body and environment which are used in parallel with the body and environment to enhance motor control and perception and to provide faster feedback during motor processes, and can be run off-line to produce imagery and evaluate sensoriomotor conterfactuals” (53).

The author agree whith the radicals on many points (see 55, 87-88):

-The emphasis in sensoriomotor control.

The Watt governor and the Timothy van Gelder interpretation

The Watt governor is a feedback controller. Many feedback controller work because they form a dynamical system. Crush calls the strong coupling thesis the claim that “brain and body/environment are coupled in the way that a controller and plant are coupled in a feedback control scheme” (59). (closed-loop control and mutual interaction).  This lead to an “radical” idea called by the author “the strong coupling thesis” (59), according to which “the brain and body/environment are coupled in the way that a controller and plant are coupled in a feedback control scheme –they are in a state of mutual causal influence, and appropriate behaviors are an outcome of this interaction” (59).  See “the dependence thesis”.

ONE PROBLEM: How identify and individualize (analyse) components? (67)

-Functions: Interface/component argument (Haugeland)

-Activities (Brook)

-Plug criterion (Gursh)

I find very interesting the idea that the brain could employ different strategies for different purposes, and from this perspective different models of the mind/body system may be right or reveals some aspect of its organization. In this sense the goal of the author would be to add the emulation framework to a set of tools consisting of theories such as classical symbol manipulation, connectionism, and dynamical system theory (88).

Finally, could be interesting to note the rhetorical or argumentative strategy followed by the author, namely, to accept some premises from those he is going to criticize leading to different conclusions:  “…Showing how even on their own terms, there is a powerful model of sensorionotor function that underwrites the Cartesian assumptions” (55). Another example is that  the concept of representation satisfices the conditions established by Haugeland in order to be a representation (see 83-84).

LA MENTE COMPUTACIONAL

En esta entrada consideraré algunas cuestiones sobre la concepción computacional de la mente, comenzando por lo que significa que una mente compute o que el cerebro-mente humano posea propiedades computacionales, una idea clave en la constitución de la Ciencia Cognitiva. Entre los coceptos que habrá que examinar las nociones de símbolo y de representación serán cruciales.

EL ANÁLISIS DE LA MEMORIA DE STERNBERG