Scientific Understanding of Consciousness
Cognition Regulation in Dorsal Anterior Cingulate Cortex
Nature 488, 218–221 (09 August 2012)
Human dorsal anterior cingulate cortex neurons mediate ongoing behavioural adaptation
Nayef Al-Rodhan Laboratories, Department of Neurosurgery, Massachusetts General Hospital, Boston, Massachusetts 02114, USA
Sameer A. Sheth, Matthew K. Mian, Shaun R. Patel, Ziv M. Williams & Emad N. Eskandar
Department of Anatomy and Neurobiology, Boston University School of Medicine, Boston, Massachusetts 02118, USA
Shaun R. Patel
Department of Neurosurgery, Alpert Medical School, Brown University and Rhode Island Hospital, Providence, Rhode Island 02912, USA
Wael F. Asaad
Department of Psychiatry, Massachusetts General Hospital, Boston, Massachusetts 02114, USA
Darin D. Dougherty & George Bush
The ability to optimize behavioural performance when confronted with continuously evolving environmental demands is a key element of human cognition. The dorsal anterior cingulate cortex (dACC), which lies on the medial surface of the frontal lobes, is important in regulating cognitive control. Hypotheses about its function include guiding reward-based decision making, monitoring for conflict between competing responses and predicting task difficulty. Precise mechanisms of dACC function remain unknown, however, because of the limited number of human neurophysiological studies. Here we use functional imaging and human single-neuron recordings to show that the firing of individual dACC neurons encodes current and recent cognitive load. We demonstrate that the modulation of current dACC activity by previous activity produces a behavioural adaptation that accelerates reactions to cues of similar difficulty to previous ones, and retards reactions to cues of different difficulty. Furthermore, this conflict adaptation, or Gratton effect, is abolished after surgically targeted ablation of the dACC. Our results demonstrate that the dACC provides a continuously updated prediction of expected cognitive demand to optimize future behavioural responses. In situations with stable cognitive demands, this signal promotes efficiency by hastening responses, but in situations with changing demands it engenders accuracy by delaying responses.
Human cognition is characterized by the ability to parse and evaluate a stream of constantly changing environmental stimuli so as to choose the most appropriate response in evolving conditions. The dACC is thought to be important in regulating cognitive control over goal-directed behaviour. Various theories postulate its involvement in linking reward-related information to action, monitoring for conflict between competing responses or detecting the likelihood of error commission. Despite substantial information from studies using lesions, functional magnetic resonance imaging (fMRI) and event-related potentials, the neurophysiological basis of its regulatory role remains the subject of considerable debate.
We studied dACC function with a combination of fMRI, single-neuronal recordings and observations of behaviour before and after lesion in human subjects undergoing surgical cingulotomy, a procedure in which a precise, stereotactically targeted lesion is created in the dACC. Microelectrode recordings, which are routinely performed during the procedure, allowed us to record from individual dACC neurons. Six subjects participated, and in four of these we also obtained a preoperative fMRI with the same task. In four we recorded behavioural responses using the same task immediately after cingulotomy.
We recorded 59 well-isolated single dACC neurons, with an average baseline firing rate of 5.7 ± 0.7 (mean ± s.e.m.) spikes per second.
Current models of dACC function, whether predicated on conflict monitoring, reinforcement learning or reward-based decision making, require that future dACC activity reflect past experience, but modulation of dACC firing on the basis of recent history has not been demonstrated at the single-neuronal level.
Our results support the view that the dACC is specifically responsible for providing a continuously updated account of predicted demand on cognitive resources. The salient influence of current dACC activity on future neuronal activity and behaviour permits the implementation of behavioural adjustments that optimize performance. In situations in which cognitive demands remain constant, this signal facilitates efficiency by accelerating responses. In situations involving rapidly changing demands, it promotes accuracy by retarding responses.
We enrolled six study subjects (four male, ages 37.5 ± 5 years (mean ± s.e.m.)) undergoing stereotactic cingulotomy for treatment-refractory obsessive–compulsive disorder. Evaluation for surgical candidacy was conducted by a multidisciplinary team consisting of psychiatrists, neurologists and neurosurgeons. Subjects enrolled voluntarily, providing informed consent under a protocol approved by the Massachusetts General Hospital Institutional Review Board. The surgical procedure produces a stereotypical lesion with an average volume of 3.58 ± 1.24 cm3 (mean ± s.d.), centred 9 mm lateral to the midline, 18 mm anterior to the anterior commissure, and 30 mm superior to the anterior-commissure–posterior-commissure plane. Subject participation was in no way related to clinical decision-making regarding their candidacy for surgery.
During the microelectrode recording portion of surgery, subjects performed the multi-source interference task (MSIT). The task was presented on a computer monitor using a customized software package in MATLAB (MathWorks). Each trial contained a stimulus consisting of three integers from 0 to 3. One number (the unique ‘target’) differed from a pair of ‘distracter’ numbers (for example 100 or 323). Subjects were asked to report, by pressing a button, the identity (rather than the position) of the target (left button for 1, middle for 2, right for 3).
Functional MRI was performed before surgery by using a 3.0-T scanner (Allegra, Siemens AG) and head coil. The MSIT was presented on a screen visible by means of a tilted mirror, and controlled with MacStim 2.6 software (WhiteAnt Occasional Publishing). Scans were acquired with the following specifications: 15 coronal sections, 64 × 64 matrix, 3.125 mm2 in-plane resolution, 5 mm thickness with 0 mm skip, 30 ms echo time, 1,500 ms repetition time, 90° flip angle, 20 cm2 field of view. The task was run in a block design. Each block consisted of 24 trials of the same type. One run consisted of eight alternating blocks, with an additional five visual fixation blocks interspersed. Data were analysed with Brain Voyager software (Brain Innovation). Anatomical and functional data were registered and transformed into common Talairach space. A general linear model was constructed by using predictors modelled by convolution with a standard haemodynamic response function. Single-subject repeated-measures ANOVAs were performed on a voxel-wise basis. Multiple comparisons were accounted for by using a cluster constraint with regional false-positive probability P < 10−4. This constraint required a cluster of at least seven contiguous voxels with P < 0.05.
The surgical procedure was performed with standard stereotactic techniques. A Cosman–Roberts–Wells (Integra) stereotactic frame was affixed to the patient under local anaesthesia, and a high-resolution MRI was obtained. The target for the left posterior lesion (2 cm posterior to the most anterior point of the frontal horn of the lateral ventricle, 0.7 cm lateral to midline, and 0.5 cm superior to the corpus callosum) was programmed into a neuro-navigation computer (StealthStation, Medtronic) and the stereotactic frame was then set appropriately. The patient was positioned semi-recumbent, the surgical area was prepared, and sterile drapes were applied. Local anaesthetic was infiltrated, a coronal skin incision was performed, and bilateral burr holes were drilled 1.5 cm lateral to the midline and 10.0 cm posterior to the nasion. A computerized microelectrode drive controlled by a neurophysiology system (Alpha Omega) was affixed to the frame. After dural opening, microelectrodes were lowered using the computerized drive in increments of 0.01 mm. The position of the tip of the electrodes was also monitored in real time using the stereotactic neuronavigation system. After microelectrode recordings, a thermocoagulation electrode with a 10-mm exposed tip (Cosman Medical) was lowered to the target. Lesions were performed by heating the electrode to 85 °C for 60 s. Two more pairs of lesions were then created, each 7 mm anterior and 2 mm inferior to the previous lesion.
For microelectrode recordings, an array of three tungsten microelectrodes (500–1,500 kΩ; FHC) was attached to a motorized microdrive (Alpha Omega Engineering). As per routine surgical protocol, recordings were obtained from the left hemisphere. On reaching the cingulate cortex, microelectrodes were held in place and monitored for about 5 min to assess signal stability. Putative neurons were not screened for task responsiveness. Analogue data were amplified, bandpass filtered between 300 Hz and 6 kHz, sampled at 20 kHz (Alpha Omega Engineering) and spike-sorted (Offline Sorter, Plexon).
[end of paraphrase]
Return to — Cognition, Emotion, Motivation