Natural Rationality | decision-making in the economy of nature


Cognitive Control and Dopamine: A Very Brief Intro

In certain situations, learned routines are not enough. When situations are too uncommon, dangerous and difficult or when they require the overcoming of a habitual response, decisions must be guided by representations. Acting upon an internal representation is referred to, in cognitive science, as cognitive control or executive function[1]. The agent is lead by a representation of a goal and will robustly readjust its behavior in order to maintain the pursuit of a goal. The behavior is then controlled ‘top-down’, not ‘bottom-up’. In the Stroop task, for instance, subject must identify the color of written words such as ‘red’, ‘blue' or ‘yellow’ printed in different colors (the word and the ink color do not match). The written word, however, primes the subject to focus on the meaning of the word instead of focusing on the ink’s color. If, for instance, the word “red” is written in yellow ink, subjects will utter “red” more readily than they say “yellow”. There is a cognitive conflict between the semantic priming induced by the word and the imperative to focus on the ink’s color. In this task, cognitive control mechanisms ought to give priority to goals in working memory (naming ink color) over external affordances (semantic priming). An extreme lack of cognitive control is exemplified in subjects who suffer from “environmental dependency syndrome”[2]: they will spontaneously do what their environment indicates of affords them: for instance, they will sit on a chair whenever they see one, or undress and get into a bed whenever they are in presence of a bed (even if it’s not in a bedroom).

Cognitive control is thought to happen mostly in the prefrontal cortex (PFC),[3] an area strongly innervated by midbrain dopaminergic fibers. Prefrontal areas activity is associated with maintenance and updating of cognitive representations of goals. Moreover, impairment of these areas results in executive control deficits (such as the environmental dependency syndrome). Since working memory is limited, however, agents cannot hold everything in their prefrontal areas. Thus the brain faces a tradeoff between attending to environmental stimuli (that may reveal rewards or danger, for instance) and maintaining representation of goals, viz. the tradeoff between rapid updating and active maintenance [4]. Efficiency requires brains to focus on relevant information and again, dopaminergic systems are involved in this process. According to many researches[5], dopaminergic activity implements a ‘gating’ mechanism, by which the PFC alternates between rapid updating and active maintenance. A higher level of dopamine in prefrontal area signals the need to rapidly update goals in working memory (rapid updating: ‘opening the gate’), while a lower level induces resistance to afferent signals and thus a focus on represented goals (active maintenance: ‘shutting the gate’). Hence dopaminergic neurons select which information (goal representation or external environment) is worth paying attention to. This mechanisms is thought to be implemented by different dopamine receptors, the D1 and D2 being responsive to different dopamine concentration (D1-low, D2-high):

Fig. 1 (From O'Reilly, 2006). Dopamine-based gating mechanism that emerges from the detailed biological model of Durstewitz, Seamans, and colleagues. The opening of the gate occurs in the dopamine D2-receptor–dominated state (State 1), in which any existing active maintenance is destabilized and the system is more responsive to inputs. The closing of the gate occurs in the D1-receptor–dominated state (State 2), which stabilizes the strongest activation pattern for robust active maintenance. D2 receptors are located synaptically and require high concentrations of dopamine and are therefore activated only during phasic dopamine bursts, which thus trigger rapid updating. D1 receptors are extrasynaptic and respond to lower concentrations, so robust maintenance is the default state of the system with normal tonic levels of dopamine firing.

Here is a neurobiological description of the phenomena, with neuroanatomical details:

Fig. 2. (From O'Reilly, 2006). Dynamic gating produced by disinhibitory circuits through the basal ganglia and frontal cortex/PFC (one of multiple parallel circuits shown). (A) In the base state (no striatum activity) and when NoGo (indirect pathway) striatum neurons are firing more than Go, the SNr (substantia nigra pars reticulata) is tonically active and inhibits excitatory loops through the basal ganglia and PFC through the thalamus. This corresponds to the gate being closed, and PFC continues to robustly maintain ongoing activity (which does not match the activity pattern in the posterior cortex, as indicated). (B) When direct pathway Go neurons in striatum fire, they inhibit the SNr and thus disinhibit the excitatory loops through the thalamus and the frontal cortex, producing a gating-like modulation that triggers the update of working memory representations in prefrontal cortex. This corresponds to the gate being open.

Hence it is interesting to note that dopaminergic neurons are involved in basic motivation and reinforcement, and in more abstract operations such as cognitive control.

Notes and references
  1. (Norman & Shallice, 1980; Shallice, 1988)
  2. (Lhermitte, 1986)
  3. (Duncan, 1986; Koechlin, Ody, & Kouneiher, 2003; Miller & Cohen, 2001; O’Reilly, 2006)
  4. (O’Reilly, 2006)
  5. (Montague, Hyman, & Cohen, 2004; O'Donnell, 2003; O’Reilly, 2006)

  • Durstewitz, D., Seamans, J. K., & Sejnowski, T. J. (2000). Dopamine-Mediated Stabilization of Delay-Period Activity in a Network Model of Prefrontal Cortex. Journal of Neurophysiology, 83(3), 1733-1750.
  • Duncan, J. (1986). Disorganization of behavior after frontal lobe damage. Cognitive Neuropsychology, 3(3), 271-290.
  • Koechlin, E., Ody, C., & Kouneiher, F. (2003). The Architecture of Cognitive Control in the Human Prefrontal Cortex. Science, 302(5648), 1181-1185.
  • Lhermitte, F. (1986). Human autonomy and the frontal lobes. Part 11: Patient behavior in complex and social situations: The “environmental dependency syndrome.” Annals of Neurology, 19(4), 335–343.
  • Miller, E. K., & Cohen, J. D. (2001). An integrative theory of prefrontal cortex function. Annual Review of Neuroscience, 24(1), 167-202.
  • Montague, P. R., Hyman, S. E., & Cohen, J. D. (2004). Computational roles for dopamine in behavioural control. Nature, 431(7010), 760.
  • Norman, D. A., & Shallice, T. (1980). Attention to Action: Willed and Automatic Control of Behavior: Center for Human Information Processing, University of California, San Diego.
  • O'Donnell, P. (2003). Dopamine gating of forebrain neural ensembles. European Journal of Neuroscience, 17(3), 429-435.
  • O’Reilly, R. C. (2006). Biologically Based Computational Models of High-Level Cognition Science, 314, 91-94.
  • Shallice, T. (1988). From neuropsychology to mental structure. Cambridge [England] ; New York: Cambridge University Press.