The Implicit Mind: Cognitive Architecture, The ...
Connectionist Learning with Adaptive Rule Induction On-line (CLARION) is a computational cognitive architecture that has been used to simulate many domains and tasks in cognitive psychology and social psychology, as well as implementing intelligent systems in artificial intelligence applications. An important feature of CLARION is the distinction between implicit and explicit processes and focusing on capturing the interaction between these two types of processes. The system was created by the research group led by Ron Sun.
The Implicit Mind: Cognitive Architecture, the ...
CLARION is an integrative cognitive architecture, it is used to explain and simulate cognitive-psychological phenomena, which could potentially lead to an unified explanation of psychological phenomena. There are three layers to the CLARION theory, the first layer is the core theory of mind. The main theories consists of a number of distinct subsystems, which are the essential structures of CLARION, with a dual representational structure in each subsystem (implicit versus explicit representations; Sun et al., 2005). Its subsystems include the action-centered subsystem, the non-action-centered subsystem, the motivational subsystem, and the meta-cognitive subsystem. The second layer consists of the computational models that implements the basic theory, it is more detailed than the first level theory but is still general. The third layer consists of the specific implemented models and simulations of the psychological processes or phenomena. The models of this layer arise from the basic theory and the general computational models.
The distinction between implicit and explicit processes is fundamental to the Clarion cognitive architecture.[1] This distinction is primarily motivated by evidence supporting implicit memory and implicit learning. Clarion captures the implicit-explicit distinction independently from the distinction between procedural memory and declarative memory. To capture the implicit-explicit distinction, Clarion postulates two parallel and interacting representational systems capturing implicit an explicit knowledge respectively. Explicit knowledge is associated with localist representation and implicit knowledge with distributed representation.
Explicit knowledge resides in the top level of the architecture, whereas implicit knowledge resides in the bottom level.[1][2] In both levels, the basic representational units are connectionist nodes, and the two levels differ with respect to the type of encoding. In the top level, knowledge is encoded using localist chunk nodes whereas, in the bottom level, knowledge is encoded in a distributed manner through collections of (micro)feature nodes. Knowledge may be encoded redundantly between the two levels and may be processed in parallel within the two levels. In the top level, information processing involves passing activations among chunk nodes by means of rules and, in the bottom level, information processing involves propagating (micro)feature activations through artificial neural networks. Top-down and bottom-up information flows are enabled by links between the two levels. Such links are established by Clarion chunks, each of which consists of a single chunk node, a collection of (micro)feature nodes, and links between the chunk node and the (micro)feature nodes. In this way a single chunk of knowledge may be expressed in both explicit (i.e., localist) and implicit (i.e., distributed) form, though such dual expression is not always required.
CLARION has been used to account for a variety of psychological data (Sun, 2002, 2016), such as the serial reaction time task, the artificial grammar learning task, the process control task, a categorical inference task, an alphabetical arithmetic task, and the Tower of Hanoi task. The serial reaction time and process control tasks are typical implicit learning tasks (mainly involving implicit reactive routines), while the Tower of Hanoi and alphabetic arithmetic are high-level cognitive skill acquisition tasks (with a significant presence of explicit processes). In addition, extensive work has been done on a complex minefield navigation task, which involves complex sequential decision-making. Work on organizational decision tasks and other social simulation tasks (e.g., Naveh and Sun, 2006), as well as meta-cognitive tasks, has also been initiated.
One problem with focusing on impact awareness, however, as Holroyd(2012) points out, is that we may be unaware of the impact of a greatmany cognitive states on our behavior. The focus on impact awarenessmay lead to a global skepticism about moral responsibility, in otherwords. This suggests that impact awareness may not serve as a goodcriterion for distinguishing responsibility for implicit biases fromresponsibility for other cognitive states, notwithstanding whetherglobal skepticism about moral responsibility is defensible.
As Payne and Gawronski (2010) note, this version of implicitness (hereafter a-implicitness) focuses the learning and cultural internalization process, isolating the relational property of acquired facility, or expertise (captured in the concept of automaticity) a given agent has gained with regard to the cultural-cognitive kind in question.
When transferred to such cultural-cognitive kinds as beliefs or attitudes, the a-implicitness criterion disaggregates into two sub-criteria. We may say of an attitude that is a-implicit if it (a) automatically activated or (b) once activated, applied or put to use in an efficient and non-resource demanding manner.
Thus, a stereotype for a category (filling in open slots in the schema with non-negotiable default) is a-implicit when its activation happens without much intervention (or control) on the part of the agent after exposure to a given environmental cue or prompt. A given stereotype may also be a-implicit in that, once activated, individuals cannot help but to use for purposes of categorization, inference, behavior, and so on. One thing that is not implied when ascribing a-implicitness is that agents are not aware of their using a cultural-cognitive kind in question. For instance, people may be very well aware that their using a default stereotype for a category (e.g., I feel this neighborhood is dangerous) even if this stereotype was automatically activated.
A branching diagram depicting the different types of implicitness discussed so far is shown in Figure 1 above. First, the notion of implicitness splits into two distinct properties, one applicable to public (non-mental) cultural kinds and the other applicable to cultural-cognitive kinds. Then this latter one splits into what I have referred to as a-implicitness and u-implicitness. A-implicitness, in turn, may refer to automaticity of activation or automaticity of application (or both) and u-implicitness may refer to unawareness of source (learning history), unawareness of the content of the cultural-cognitive kind itself when it is operating (e.g., an unconscious attitude, belief, schema, etc.), or unawareness that the activation of this cultural-cognitive kind influences action.
While these issues are too complex to deal with here, the conceptual cautionary tale is that it is better to be explicit and granular about implicitness, especially when ascribing this property to a cultural-cognitive element as part of the explanation of how that element links to action.
Finally, the Clarion architecture is a hybrid cognitive architecture with both connectionist and symbolic representations, that combines implicit and explicit psychological processes, and integrates cognition (in the narrow sense) and other psychological processes. Overall, Clarion is a modularly structured cognitive architecture consisting of a number of distinct subsystems, with a dual representational structure in each subsystem [52].
Schizophrenia is characterized by impairments in several domains of social cognition, including Theory of Mind (ToM) or mentalizing1,2,3 and goal or intention attribution4,5,6. Mentalizing deficits have a clear functional importance in schizophrenia, as they are one of the strongest predictors of functional outcome among the other social as well as non-social cognitive domains7. Overall, research on social cognition in schizophrenia leaves some questions open, including three that will be our main focus here: (1) Do individuals with schizophrenia show a hypo- or a hyper-mentalizing deficit? (2) Is this deficit characterized by an impairment in the attribution of intention or contingency? (3) Is this deficit situated at an implicit and automatic or an explicit and reflexive level of processing? Hypomentalizing refers to being less able to perceive and infer goals and intentions. In contrast, hypermentalizing would involve over-attributing intentions, including to non-intentional stimuli. Hypermentalizing has been suggested by several authors by the existence of paranoid symptoms in schizophrenia, leading to an excessive attribution of malevolent intentions to others8,9. This hypothesis has received some experimental evidence at the behavioral10,11,12,13 and the neurophysiological levels14,15,16. Nevertheless, replications of this result are inconsistent. Several studies have reported hypomentalizing without hypermentalizing in schizophrenia17,18,19,20. In another study, hypermentalizing in schizophrenia was entirely explained by lower verbal intelligence and memory, which was not the case for hypomentalizing21. It has also been suggested that hypermentalizing may be related to apophenia, i.e. an increased tendency to perceive connections between unrelated events22. However, only one study reported that patients with delusions of persecution over-attributed contingency, whereas no differences were found with control or non-deluded groups for the attribution of intentions23. The issue of hypo- vs. hyper-mentalizing in schizophrenia thus remains largely open.
One limitation to the present study came from the fact that the paradigm was not entirely implicit, as participants were required to focus their attention on the animations because they would be asked to describe them. In an entirely implicit design, participants would have been asked to perform a task unrelated to the description of animations, while eye movements were recorded. This study does not explore the stages between the visual processing of triangles and the interpretation of the scene. Another limitation was related to the DSM-IV-R diagnosis interviews of patients and controls that did not involve a structured instrument like the Structured Clinical Interview for Disorders. There was a discrepancy between the results on the length scale and the results on the contingency/intentionality scale. Although descriptions were as long in patients and controls, patients described fewer intentional actions in GD and ToM animations than controls. This decrease was not compensated by an increase in the description of the mechanical movements in these two conditions. There was a global reduction in the number of actions described in these two conditions: patients might have committed more repetitions, or described the triangles more on their physical appearance rather on their actions. The present results could thus also be interpreted according to the experiential (perceptual) vs. narrative dichotomy, as patients were impaired in the verbal description of GD and ToM animations. Indeed, schizophrenia has been characterized by an uninformative speech with a low idea density57: our results suggest that this deficit seems to occur only when they have to infer simple and complex intentions, but not mechanical actions. Finally, the paradigm used two different and heterogeneous conditions of mentalizing. The first condition (GD) implies a basic form of mentalizing, with the attribution of simple intentions to others. The second condition (ToM) involves a complex form of mentalizing, with the attribution of complex intentions to others. It is worth noting that differences in intentionality between patients and controls were partly explained by contextual control difference for ToM but not for GD: complex mentalizing may require complex cognitive processing like the integration of different sources of contextual information whereas basic mentalizing may rely more on perceptual information. The final limitation relates to the impact of the kinematic differences on the interpretation of eye tracking differences between GD and ToM. In a previous study, we had demonstrated that kinematic confounds did not explain eye movement differences between R and GD and between R and ToM. However, eye movement differences between GD and ToM were entirely explained by kinematic confounds. Thus our study validated the dissociation between explicit mentalizing and implicit attribution of intentions in general: however, it was not possible to compare the basic implicit mentalizing with the complex explicit one. 041b061a72