The following is a brief anatomical basis of attention. We are not looking for an exhaustive bibliographical or theoretical exhaustiveness of the concept, only a brief exposition of the main nodes involved in the attentional process, and how they do it. We will start from Posner’s model (which is the most widely accepted) and implement it with current research.
Attention is “the selection of information for conscious processing and action, as well as the maintenance of the alertness required for attentional processing” (Posner and Bourke, 1999).
Fundamental concepts of attention
Posner (1995) points out three fundamental concepts of attention.
- Attention does not process information; it merely enables or inhibits that processing. Attention can be differentiated anatomically from information processing systems.
- Attention is supported by anatomical networks; it does not belong to a specific area of the brain, nor is it a global product of the brain.
- The brain areas involved in attention do not have the same function; rather, different functions are supported by different areas. It is not a unitary function.
Esto es muy importante ya que permite diferenciar el sistema de procesamiento atencional del sistema de This is very important because it allows differentiating the attentional processing system from the perceptual processing system, or others, which is very relevant from the evaluative point of view. Attention, therefore, has two main functions: maintaining the state of alertness (VIGILANCE) and selecting the information to which resources will be devoted (MONITORING AND CONTROL). It selects the mechanisms and information to be manipulated. The attentional system has a limited capacity, hence the need to select the type of relevant information.
Posner distingue tres sistemas atencionales:
- Ascending reticular system: In charge of tonicity tasks, regulation of wake states and autonomic state for functioning. Its main nuclei are located in the brain stem, although its networks extend through the ascending pathways throughout the brain. Its main neurotransmitter is norepinephrine (NE). The main inputs of NE from the locus coeruleus are the parietal area, the pulvinar nucleus of the thalamus and the colliculi, i.e., the areas that form the posterior attentional network.
- Posterior attentional system: This network is related to the visuospatial orientation of attention, which is why it has also been called orientation network. The term orientation refers to the overt alignment (of the sensory organs) or covert alignment (of the attention) with a source of sensory information or with a memory content. Orientation can be exogenous or respond to internal elements.
- Anterior attentional system: Performs functions of stimulus selection and discrimination and error detection.
The activity of the systems is not independent per se, but depends on the attentional demands of the task.
Attentional system in Posner’s theory
For Benedet (2002), the attentional system in Posner’s theory implies:
- Maintaining the necessary state of alertness at each moment.
- Detecting infrequent stimulus changes (vigilance).
- Selecting relevant information (selective attention) and inhibiting irrelevant information (resistance to distraction).
- Maintain this selective function during the execution of an activity or task of a certain duration (sustained attention),
- Evaluate the state of the system at any given moment (monitoring);
- To optimally distribute resources among the different representations and operations that are activated (distributed attention).
Improving Posner’s model
Some authors, although they partially agree with Posner, implement his attentional model. Dosenbach, Fair, Cohen, Schlaggar & Petersen (2008) say that in the study of attention and especially of the attentional control network, there are several methods of analysis to draw conclusions about the connectivity of attentional networks. These methods limit the extent of conclusions whether they are used separately or together.
They do not negate Posner’s initial proposal, as there are a number of attentional networks with distinct purposes; although their results support a redefinition of the attentional control system. To this end, Dosenbach et al. (2008) study top-down phenomena through a combination of different techniques, in a theoretical framework based on the theory of complex systems.
This theory proposes that there are a number of nodes in the central nervous system that are interconnected in an efficient manner, such that they produce “small-world” neural architectures, in which the nodes of the system are neither random nor regular.
First, they propose a mixed block- and event-related design, which provides a much more refined measure of the various processes of monitoring: attentional control initiation, cognitive set maintenance, and error detection. In this way, the analysis of activations at resonance is much finer, since it unlinks activations in different experimental situations.
Cognitive activation networks
The results suggest two cognitive activation networks: a resting state network and a cognitive activation network. This proposal has been made by Dosenbach himself, but also by other authors (Raichle et al., 2001; Corbetta et al., 2008), who have carried out an analysis of these two networks. As we said in another post, it seems that in some cases they show an inverse activation relationship: the higher the activity of the executive network, the lower the activity of the resting network.
Dosenbach et al. (2008) use other methods that allow, from the complex systems model, to analyze the dynamics of attentional control networks. Firstly, graph theory, a branch of mathematics that allows analysis between networks or nodes of different networks. In the case of neuroscience and neuropsychology, this graph theory is applied to two types of data: the ROIs (regions of interest), which would function as the nodes, and the correlations in the activation pattern of the different ROIs.
On the other hand, they analyze the directionality of the correlations in the activation pattern of the ROIs in the MRI using a technique called PPI (psychophysiological interaction analysis). This technique examines the specific context-dependent relationship at the two nodes on an acquisition basis between different trials. Together, these data provide a more specific distribution and definition of the control network(s).
Dosenbach’s experiments analyze areas that support three types of functions:
- Set change or set maintenance-attention to the signal-which involves the anterior insula, dorsal CCA, and anterior prefrontal cortex;
- Adjustment and feedback, involving the dorsolateral prefrontal cortex and inferior parietal lobe; and
- Cognitive control initiation, involving the intraparietal sulcus and dorsolateral frontal cortex.
In addition to the methods of relationship analysis, we also seek to establish the independence of the systems, for which we proceed to an independent component analysis.
Fronto-parietal and cingulo-opercular networks
The results of Dosenbach et al. (2008) show a somewhat different and more complex distribution than that of Posner. There are two attentional control networks, one fronto-parietal and one cingulo-opercular. Both are linked through a structure that, in recent years, has begun to gain weight in research as a complex cognitive processor: the cerebellum.
The fronto-parietal network is formed by the dorsolateral prefrontal cortex, the inferior parietal lobe, the dorsal frontal cortex, the intraparietal sulcus, the precuneus, and the medial cingulate cortex. Its main function is to initiate and adjust cognitive control, responding differentially according to the feedback it receives from performance – correct vs. erroneous trials.
On the other hand, the cingulo-opercular network is formed by the anterior prefrontal cortex, anterior insula, dorsal CCA and thalamus. Its main function is to keep the cognitive set stable during task performance.
What is the function of the cerebellum and why does it exhibit significant activity? Some authors (Allen, Buxton, Wong, and Courchesne; 1997) have proposed that the cerebellum is a critical center for predicting and preparing for impending information acquisition, analysis, or action. In this double control network, the cerebellum functions as a “way station” between the thalamus (cingulo-opercular) and the precuneus, inferior parietal cortex and dorsolateral prefrontal cortex (fronto-parietal), acting as an error analysis mechanism and connecting with areas that detect and adopt strategies in the face of error.
Taken from : http://scienceblogs.com/developingintelligence/2010/09/15/machines-learn-how-humans-lear/
The main virtue of the model proposed by Dosenbach’s group is that this processing is performed in parallel, with two neural networks that process task-relevant information but whose top-down control is different.
On the one hand, the fronto-parietal network would process information relevant to adaptive control, actively keeping task-relevant information from a limited number of trials in mind, in order to implement fast control of [priority] parameter adjustments when an error is detected. Whereas the cingulo-opercular network, which involves maintenance of the cognitive set, involves sustained activity across different trials, which detects the error early on, but does not result in a change of parameters in task execution.
This finding is consistent with the proposal of Corbetta and Shulman (2002) and Corbetta et al. (2008), who establish a double attentional execution network: a ventral one, in charge of detecting the salience of environmental stimuli, and a dorsal one, which is activated in focused attention tasks with a prolonged duration, and which also acts guided by the ventral network.
However, both networks are not directly related, but are interrelated through the prefrontal cortex. In the case of Dosenbach and his group, we could extrapolate the fronto-parietal control network partially to structures of the dorsal attention network, while we could make an analogy between the detection of environmental salience (ventral) and the detection of an error during task execution.
In short, Posner’s model that includes the CCA as part of an attentional control system was not entirely complete, and the models of Dosenbach and Corbetta implement it, proposing more complex systems, involving a greater interrelation between large-scale networks in a “small-world” architecture, and an attentional control that does not depend almost exclusively on prefrontal structures.
In this sense, the CCA -and, especially, the dorsal CCA-, seems to be a processing node that gains weight over prefrontal structures as a necessary structure to perform high-level cognitive tasks; especially when it is necessary to maintain a cognitive set, and to detect that this set is failing in the execution of a task.
In fact, von Economo neurons have developed later than others that form neural structures involving adaptive functions. That is, thanks to these neurons, our goal-directed behavior is more dilated, compared to other animals, which, ontogenetically, can also be observed with respect to the evolutionary development of humans.
- Allen G1, Buxton RB, Wong EC, Courchesne E (1997). Attentional activation of the cerebellum independent of motor involvement.
- Corbetta M1, Shulman GL (2002). Control of goal-directed and stimulus-driven attention in the brain
- Dosenbach, N. U. F., Fair, D., Cohen, A. L., Schlaggar, B. L., & Petersen, S. E. (2008). A dual-networks architecture of top-down control. Trends in Cognitive Sciences, 12(3), 99-105. https://doi.org/10.1016/j.tics.2008.01.001
- Posner, M. I. (1995). Attention in cognitive neuroscience: An overview. In M. S. Gazzaniga (Ed.)
- Posner, M.I y Bourke. P. (1999): «Attention».
- Raichle ME (2001). A default mode of brain function