Decoding sequential stages of task preparation in the human brain
Introduction
What makes us wait in front of traffic lights? We follow a simple rule: We respond to the red light by stopping and to the green light by walking. In most cases, however, our behaviour requires reacting to stimuli more flexibly because rules are not always as fixed as in the case of traffic lights. Crossing the street in England, for example, requires first looking to the right, whereas in Germany it requires looking left. Hence, responses to stimuli are often context-dependent, requiring a switch between multiple motor behaviours associated with the same stimulus. This ability to act according to contextual rules requires flexible reconfiguration of the flow of information through the brain. Perceptual information that enters the primary sensory areas needs to be routed into different motor responses depending on current tasks and other contexts of behaviour. Thus, three types of representations coordinate rule-guided behaviour: the representation of a sensory stimulus needs to be transformed to the representation of a motor response, guided by a rule that is also represented in the brain. It has been suggested that the lateral prefrontal cortex (PFC) is involved in guiding flexible behaviour (Miller and Cohen, 2001). Single cell recordings in monkeys have demonstrated that a fronto-parietal network, including PFC and intraparietal sulcus (IPS), directs the translation of perceptual stimuli into different actions (Quintana and Fuster, 1999, Asaad et al., 2000, Wallis et al., 2001, Wallis and Miller, 2003, Amemori and Sawaguchi, 2006, Johnston et al., 2007). In humans, several brain areas have been shown to be activated during preparation of externally cued tasks, including lateral PFC, IPS, the pre-supplementary motor areas (pre-SMA) and the anterior cingulate cortex (ACC) (Toni et al., 1999, Dove et al., 2000, Sohn et al., 2000, Kimberg et al., 2000, Passingham et al., 2000, Bunge et al., 2002, Brass and von Cramon, 2002, Brass and von Cramon, 2004a, Brass and von Cramon, 2004b, Rushworth et al., 2001, Rushworth et al., 2002a, Rushworth et al., 2002b, Brass et al., 2003, Braver et al., 2003, Forstmann et al., 2005, Dosenbach et al., 2006, Crone et al., 2006, Yeung et al., 2006, Rowe et al., 2008). However, it has remained unclear which brain areas correspond to which processing stage reaching from stimulus representation to response execution. In particular, there is much debate regarding the roles of different prefrontal and parietal areas in time, i.e., from the retrieval of a general rule, its linkage to response representations to the final motor preparation (Braver et al., 2003, Bunge, 2004, Brass and von Cramon, 2004a, Brass and von Cramon, 2004b, Brass et al., 2005a, Bunge et al., 2005). It has also remained unclear whether brain activity related to task switching found in previous studies reflects changes in representations that encode the chosen task, or whether it instead reflects unspecific or global processes that do not differentiate between tasks.
The development of new decoding techniques for fMRI (Haynes and Rees, 2006, Norman et al., 2006) provides a powerful tool for revealing the information encoded in different brain areas. These techniques are based on the notion that a potential selectivity of neurons for different types of information, such as different stimuli or task rules, could be revealed by investigating how accurately they can be decoded from patterns of activity in a given brain region. The approach has been successfully applied to analyse representations of visual images in occipital and temporal cortex (Haxby et al., 2001, Cox and Savoy, 2003, Haynes and Rees, 2005a, Haynes and Rees, 2005b, Kamitani and Tong, 2005, Polyn et al., 2005, Williams et al., 2007), decisions (Hampton and O'Doherty, 2007, Pessoa and Padmala, 2007, Li et al., 2007) as well as self-generated intentions encoded in prefrontal cortex (Haynes et al., 2007). Additionally, multivariate decoding can be used to track how decisions build up across multiple brain regions over time (Soon et al., 2008).
In the current fMRI study, we wanted to investigate how information about task sets builds up across time when subjects are cued to change the rules that guide their behaviour. Specifically, we used multivariate decoding (Haynes and Rees, 2006, Norman et al., 2006) to disentangle the encoding of different types of information in the brain related to executive control: the encoding of the task-sets, the encoding of target stimuli and the encoding of motor responses (see also Toni et al., 1999, Passingham et al., 2000). For this, we used a task whereby subjects had to alternate between two task-sets based on a cue presented prior to each trial. The tasks required two different mappings of visual stimuli to motor responses. Using this procedure, it was possible to identify different stages of information processing in the human brain.
Section snippets
Participants
Seven male and seven female subjects took part in the study and gave written informed consent to the test procedure. The experiment was approved by the local ethics committee and was conducted according to the Declaration of Helsinki. All subjects were right-handed and had normal or corrected to normal visual acuity. Data from two subjects were excluded later due to poor task performance in one case and technical problems in recording responses in the other case. The final sample consisted of
Results
Subjects performed well and concentrated on the task with a low error rate of 2.6% (SD 5.0; range 0 to 5.9% of 64 trials per run). Only trials with correct responses were included in the following analyses.
We investigated whether distinct stages of information processing could be decoded from brain responses. Specifically, we applied multivariate pattern classification to each area in turn in order to assess whether it was possible to decode which of the two task-sets, target stimuli and motor
Discussion
Our results using multivariate classification provide new insights into the information flow through the brain during the preparation of rule-guided behaviour. We demonstrated a shift of task-set information encoded in the brain over time. First, with the presentation of one of two visually distinct task cues, their identity could be decoded from visual cortex. After the cue presentation, the information about task-sets shifted to different brain regions. The left IPS showed an early, transient
Acknowledgments
We thank Richard E. Passingham and Marcel Brass for their helpful comments on the manuscript. We would also like to thank Stefan Zysset, Carlo Reverberi, Chun Siong Soon and Anna He for assistance and Sven Gutekunst for his help with constructing the joystick and designing the MR-compatible table used in the experiment.
References (70)
- et al.
Neural activity in the primate prefrontal cortex during associative learning
Neuron
(1998) - et al.
Multiple components of lateral posterior parietal activation associated with cognitive set shifting
NeuroImage
(2005) - et al.
Left ventrolateral prefrontal cortex and the cognitive control of memory
Neuropsychologia
(2007) - et al.
When the same response has different meanings: recoding the response meaning in the lateral prefrontal cortex
NeuroImage
(2003) - et al.
The role of the inferior frontal junction area in cognitive control
Trends Cogn. Sci.
(2005) - et al.
Mechanisms of transient and sustained cognitive control during task switching
Neuron
(2003) - et al.
Dissociable contributions of prefrontal and parietal cortices to response selection
NeuroImage
(2002) - et al.
Functional magnetic resonance imaging fMRI “brain reading”: detecting and classifying distributed patterns of fMRI activity in human visual cortex
NeuroImage
(2003) - et al.
Retrieving rules for behavior from long-term memory
NeuroImage
(2005) - et al.
A core system for the implementation of task sets
Neuron
(2006)
Prefrontal cortex activation in task switching: an event-related fMRI study
Brain Res. Cogn. Brain Res.
Tracking cognitive processes with functional MRI mental chronometry
Curr. Opin. Neurobiol.
Internally generated and directly cued task sets: an investigation with fMRI
Neuropsychologia
From thought to action: the parietal cortex as a bridge between perception, action, and cognition
Neuron
Predicting the stream of consciousness from activity in human visual cortex
Curr. Biol.
Reading hidden intentions in the human brain
Curr. Biol.
Regional differences in the refractory period of the hemodynamic response: an event-related fMRI study
NeuroImage
Top-down control-signal dynamics in anterior cingulate and prefrontal cortex neurons following task switching
Neuron
Modulation of task-related neural activity in task-switching: an fMRI study
Brain Res. Cogn. Brain Res.
Beyond mind-reading: multi-voxel pattern analysis of fMRI data
Trends Cogn. Sci.
Single neurons in posterior parietal cortex of monkeys encode cognitive set
Neuron
Rule-dependent shifting of sensorimotor representation in the primate prefrontal cortex
Eur. J. Neurosci.
Task-specific neural activity in the primate prefrontal cortex
J. Neurophysiol.
Frontal lobe mechanisms that resolve proactive interference
Cereb. Cortex
The role of the frontal cortex in task preparation
Cereb. Cortex
Decomposing components of task preparation with functional magnetic resonance imaging
J. Cogn. Neurosci.
Selection for cognitive control: a functional magnetic resonance imaging study on the selection of task-relevant information
J. Neurosci.
Who comes first? The role of the prefrontal and parietal cortex in cognitive control
J. Cogn. Neurosci.
How we use rules to select actions: a review of evidence from cognitive neuroscience
Cogn. Affect. Behav. Neurosci.
Neural circuits subserving the retrieval and maintenance of abstract rules
J. Neurophysiol.
Neural circuitry underlying rule use in humans and nonhuman primates
J. Neurosci.
Top-down versus bottom-up control of attention in the prefrontal and posterior parietal cortices
Science
Neural evidence for dissociable components of task-switching
Cereb. Cortex
Involvement of the inferior frontal junction in cognitive control: meta-analyses of switching and Stroop studies
Hum. Brain Mapp.
An adaptive coding model of neural function in prefrontal cortex
Nat. Rev., Neurosci.
Cited by (145)
Exploring the effects of EEG signals on collision cases happening in the process of young drivers’ braking
2021, Transportation Research Part F: Traffic Psychology and BehaviourTemporal embedding and spatiotemporal feature selection boost multi-voxel pattern analysis decodingaccuracy
2020, Journal of Neuroscience MethodsNeural representations of social valence bias economic interpersonal choices
2020, NeuropsychologiaNew insights into the neural basis of cognitive control: An event-related fMRI study of task selection processes
2020, International Journal of PsychophysiologyCitation Excerpt :More specifically, when switching trials were compared to repeat trials, greater activation was found in networks that included: inferior parietal lobule (IPL), inferior frontal gyrus (IFG), supplementary motor area (SMA), and premotor cortex (Yeung et al., 2006; Kim et al., 2011; De Baene et al., 2012a,b; Yin et al., 2018). The IPL has been associated with attention and preparation processes (Bode and Haynes, 2009; Chiu and Yantis, 2009), the IFG with maintaining tasks (Barber and Carter, 2005), and the SMA with specific response selection and initiation (Hyafil et al., 2009; Shi et al., 2010). In addition, several switching studies have reported additional areas including the anterior cingulate cortex (ACC) (Yeung et al., 2006; Hyafil et al., 2009; De Baene et al., 2012a; Yin et al., 2018), associated with adjustment of selection processes (Woodward et al., 2008); the prefrontal cortex (Hyafil et al., 2009; Jamadar et al., 2010; Yin et al., 2018), with response inhibition (Jamadar et al., 2010); and the precuneus (Barber and Carter, 2005; Yeung et al., 2006; Kim et al., 2011; De Baene et al., 2012b) and precentral gyrus (Barber and Carter, 2005; De Baene et al., 2012b).