A preview of this full-text is provided by Springer Nature.
Content available from Nature Human Behaviour
This content is subject to copyright. Terms and conditions apply.
Articles
https://doi.org/10.1038/s41562-019-0801-5
1Donders Institute for Brain Cognition and Behaviour, Radboud University, Nijmegen, the Netherlands. 2Department of Experimental Psychology, Ghent
University, Ghent, Belgium. 3Center for Complex Systems & Brain Sciences, Florida Atlantic University, Boca Raton, FL, USA. *e-mail: walexander@fau.edu
Activity in the dorsal anterior cingulate cortex (dACC)
and surrounding regions in the medial prefrontal cortex
(mPFC) is routinely observed in neuroimaging studies of
cognitive control and decision-making1. Consequently, a number of
theoretical and computational accounts have been developed in the
past two decades to describe the role and function of dACC in cog-
nitive control2. Generally, cognitive control entails the need to sup-
press an incorrect, prepotent response to generate a correct, but less
automatic, response. Early computational models of dACC function
therefore principally addressed tasks thought to involve response
selection and inhibition, assigning the region roles in signalling
behavioural error, detecting and resolving response conflict3, select-
ing appropriate motor responses4 or predicting the likelihood of an
incorrect response5.
Although cognitive control research has traditionally focused on
response inhibition, recent work has highlighted the interactions
between control and motivation2,6. Motivation refers to the drive
to pursue specific behavioural goals to obtain desired outcomes
(such as rewards (extrinsic motivation) or other states perceived as
rewarding (intrinsic motivation)). This line of work defines motiva-
tion as the “invigorating impact, on both behaviour and cognition,
of prospective reward (both extrinsic reward such as money and
instrinsic reward tied to the satisfaction of self-relevant behavioural
goals”7. Here, we refer to motivated control as the process promot-
ing successful selection and invigoration of a behavioural response
leading to a valuable desired outcome. In this view, exerting con-
trol is costly, but also valuable, as it allows the securing of a pro-
spective reward8,9. Individuals consistently tend to avoid exerting
mental effort when possible10, and preparing for a cognitively effort-
demanding task is associated with increased dACC activity11,12.
Interestingly, activity in the same region appears to correlate with
the expectation of higher reward following task completion13. The
overlap of cognitive effort and reward signals within dACC has led
to development of new accounts of the region’s function, assigning
it a role in computing benefits and costs of actions, and integrating
these in a ‘net-value’ driving adaptive behavioural selection insitu-
ations involving exertion of cognitive control or physical effort9.
In line with these findings, an influential theoretical framework
has been proposed, the expected value of control account (EVC;
ref. 14). The EVC posits that activity in dACC reflects ‘expected
value of control’—a trade-off between cost and benefits resulting in
the selection of an optimal control signal.
In parallel, a growing amount of evidence supports a key role of
the dACC in tracking the likelihood of events (such as responses
and outcomes given a certain stimulus), and computing the dis-
crepancy between predicted and actual events (that is, prediction
errors), formalized in the predicted response outcome model (PRO;
ref. 15), which posits that the dACC signal reflects an ongoing com-
parison between expected and observed events: any unexpected,
and therefore ‘surprising’, event will produce increased activity in
dACC, and this signal contributes to updating of future predictions.
This type of surprise could be termed epistemic control: dACC
activity reflects predictive signals and error signals when predic-
tions are not met. These signals may trigger behavioural adaptation
when necessary16–20. This line of work successfully explains classical
inhibitory control effects as a function of likelihood of responses
and outcomes (errors, incongruent options and non-prepotent
responses are generally less likely, and therefore surprising). More
recently, this approach has also been applied to motivated control,
suggesting that motivationally relevant variables (for example, effort
requirements or potential reward amounts) may be monitored in a
similar fashion. In this framework, deciding to engage in effortful
behaviour is generally less likely and therefore the choice to engage
is associated with greater dACC activity (PRO–effort)21. Even in the
absence of subsequent invigoration, the choice itself to accept more
effortful tasks is infrequent (humans are generally effort-avoidant),
and would therefore elicit increased dACC activity (which, in epis-
temic control terms, reflects the likelihood of engaging in a task
given its specific effort and reward properties). The PRO–effort
proposal outlines how a likelihood-monitoring account of dACC
function holds potential for generalization to motivated invigora-
tion of behaviour. However, this account is yet to be tested against
the wide array of effort-related effects on behaviour and brain activ-
ity in the mPFC–dACC, especially considering that previous work
Surprise, value and control in anterior cingulate
cortex during speeded decision-making
Eliana Vassena 1,2, James Deraeve2 and William H. Alexander 2,3*
Activity in the dorsal anterior cingulate cortex (dACC) is observed across a variety of contexts, and its function remains intensely
debated in the field of cognitive neuroscience. While traditional views emphasize its role in inhibitory control (suppressing
prepotent, incorrect actions), recent proposals suggest a more active role in motivated control (invigorating actions to obtain
rewards). Lagging behind empirical findings, formal models of dACC function primarily focus on inhibitory control, highlight-
ing surprise, choice difficulty and value of control as key computations. Although successful in explaining dACC involvement in
inhibitory control, it remains unclear whether these mechanisms generalize to motivated control. In this study, we derive pre-
dictions from three prominent accounts of dACC and test these with functional magnetic resonance imaging during value-based
decision-making under time pressure. We find that the single mechanism of surprise best accounts for activity in dACC during
a task requiring response invigoration, suggesting surprise signalling as a shared driver of inhibitory and motivated control.
NATURE HUMAN BEHAVIOUR | VOL 4 | APRIL 2020 | 412–422 | www.nature.com/nathumbehav
412
Content courtesy of Springer Nature, terms of use apply. Rights reserved
