Do EEG paradigms work in fMRI? Varying task demands in the visual oddball paradigm: Implications for task design and results interpretation
Introduction
The oddball paradigm is one of the most frequently used tasks to assess target detection capabilities in cognitive neuroscience. The paradigm has its origins in electroencephalography (EEG) research where it was developed for investigating the P300 component of the event related potential (ERP) (Squires et al., 1975). The P300 represents target detection and is elicited upon recognition of an infrequent target stimulus. In addition to being a robust component that is useful for investigating aspects of cognitive function, the P300 is also modulated by certain disease states or physiological conditions. For example, a reduction in P300 amplitude is seen in patients with schizophrenia and their relatives (Turetsky et al., 2007), heavy smokers (Mobascher et al., 2010) and the ageing population (Li et al., 2012). Given the utility of the oddball paradigm (and associated P300) for fundamental neuroscience and translational medicine approaches, its transfer to other imaging modalities was inevitable. In particular, oddball tasks have been drawn on for functional magnetic resonance imaging (fMRI) due to its higher spatial resolution (down to millimetre resolution), non-invasiveness and widespread availability.
In fMRI the oddball task is known to elicit a blood oxygen level dependent (BOLD) response in a large distributed network including the supramarginal gyrus, frontal cortex, insula, thalamus, cerebellum, occipital–temporal, superior temporal and cingulate regions (Bledowski et al., 2004, Gur et al., 2007, Kiehl et al., 2005, Musso et al., 2006, Strobel et al., 2008, Warbrick et al., 2009). This is consistent with what is known about the multiple generators of the P300 component (Halgren et al., 1995a, Halgren et al., 1995b, Halgren et al., 1998). However, this widespread activation makes it difficult to pinpoint specific effects of task manipulations or group differences on brain activation and associated cognitive function. This process of assigning cognitive functions to brain activation measures is often easier in the EEG domain where the paradigm has its origin. For example, the amplitude of the P300 component is known to be indicative of the amount of attentional resources allocated to a stimulus (Picton, 1992, Pritchard et al., 2004). While many cognitive processes contribute to generating the component (e.g., attention, working memory), the single electrophysiological marker (although the P300 can be considered to consist of the P3a and P3b subcomponents, for a review of see Polich and Criado, 2006) makes interpreting modulation of EEG oddball data relatively straightforward. The amplitude and latency of the P300 can be modulated by paradigm manipulations (such as frequency of stimuli, inter-trial interval, type of response required) and the temporal resolution of EEG data means that sensory and cognitive processes can be separated on a millisecond timescale. The network of brain regions ‘lighting up’ in typical oddball fMRI measurements however, is known to subserve multiple cognitive functions as demonstrated by the vast literature examining numerous mental processes in the fMRI literature. For example, regions involved in attention allocation, decision making, working memory and motor preparation/execution are all involved in the oddball task. However, we cannot be certain whether they are related to the specific process under investigation (target detection/perceptual decision making) or whether their involvement stems from other task demands such as ongoing working memory processes or motor preparation. In short, cognitive neuroscience has developed an experimental task optimally suited to probe target detection in EEG. A transfer of the oddball task into the fMRI environment, based on the above considerations alone, is not (yet) necessarily feasible.
The use of the oddball paradigm in fMRI studies also introduces further methodological issues. For example, subjects are usually required to indicate the recognition of target stimuli by pressing a response button, thus allowing accurate recording of correct/incorrect responses and reaction time. While EEG recordings are susceptible to artefacts as a result of the subject moving, fMRI data are influenced to a greater extent, with head motion resulting from motor responses impacting on image quality. Perhaps more problematic however, is that the motor response for the traditional button press oddball paradigm could impact on stimulus locked fMRI activation and the subsequent interpretation of that activation. This has led to the use of tasks employing mental strategies as indirect measures of target detection, which do not rely on motor responses. For instance, counting the stimuli has been proposed (Kirino et al., 2000) as a means of maintaining attention on the task and ensuring task compliance in fMRI. Alternatively a truly/fully passive task design (from the subject's perspective) can be used whereby the novelty of a target stimulus is assumed to elicit a target detection-like response in the EEG, thus totally eliminating the need for a motor response. The variations in the instructions given and the response required from the subject will lead to different patterns of activation resulting from different cognitive processes associated with the changing cognitive demands associated with changing requirements. Given the multiple regions involved in responses to the oddball task, understanding these effects is crucial to the interpretation of fMRI studies using the oddball task. However, only few prior fMRI studies have investigated the effects of response modality variation in the oddball task and the findings are inconclusive. Linden et al. (1999) for example, found no response modality specific activation when comparing visual and auditory count and respond conditions in an early fMRI study. However, the instructions used for the two conditions differed on more than presence and absence of a motor response, making comparison difficult. Thus, the influence of response modality on the BOLD activation maps in the visual oddball task remains unclear.
The concerns regarding the use of motor responses in fMRI studies are not limited to the oddball paradigm. In many cases a balance needs to be struck between the need for an overt behavioural measure and reduction in motion artefacts. This is particularly true for paradigms where the overt behaviour is not necessarily the primary outcome measure and is rather used to maintain attention on the task or to ensure compliance with the task instructions. This raises two questions: 1. Is it possible to achieve the same (or at least comparable) neural activation in brain regions related to cognitive aspects of the task using a simple measure of task compliance rather than a trial-by-trial response?; 2. Does the motor response impact on the fMRI data in a negative way? Two main concerns relate to the second question. First, brain activation associated with the preparation and execution of a motor response may itself make it difficult to identify and interpret brain activation associated with the pure cognitive processes under investigation (e.g. target detection), for instance through wide spread areas of motor-related activation in parts overlapping with ‘cognitive activations’. Second, an overt motor response does not necessarily mirror those cognitive processes underlying the cognitive construct of interest.
The present study therefore uses the well-established oddball paradigm to illustrate methodological and data interpretation-related pitfalls to be considered and overcome when drawing on experimental paradigms from other imaging domains. In particular, we try to tease apart whether, or under which circumstances, neuroscientists can assume that a subject is performing a given task correctly in the absence of a motor response. Conversely, this study aims to separate processes involved in cognitive target detection and associated motor responses by dissociating cognitive from motor-response related brain activation.
In summary, while the oddball paradigm has received considerable attention in the electrophysiology literature and is a robust and reliable marker of cognitive function (Picton, 1992, Polich and Criado, 2006), thorough exploration of the paradigm in fMRI is lacking. Examining the impact of response modality on the resulting BOLD activation maps has implications not only for the oddball paradigm but also other paradigms where an overt or covert response could be used. In the present study we investigate the effects of task demands on the BOLD activation in response to a visual oddball task and consider the implications for task design in fMRI studies by answering the following questions:
- 1.
Under which circumstances is a trial-by-trial behavioural response required to measure brain activation in response to target stimuli in an oddball task?
- 2.
How does the motor response to target stimuli impact on the interpretation of BOLD activation in relation to the cognitive demands of the task?
- 3.
Why does the response modality influence the conclusions we can draw regarding brain function during the visual oddball task?
Section snippets
Methods
The study protocol was approved by the local Human Subjects Review Board at RWTH Aachen University and was carried out in accordance with the Declaration of Helsinki. Subjects were recruited via flyers, word of mouth and newsletter alerts.
Behavioural data
Mean reaction time in the respond condition was 457 ms (SD = 56) with subjects correctly responding to 38–40 (mean = 40, SD = .5) of the target stimuli. Only 1–2 (mean = 1, SD = .8) false alarms were made in the respond condition. In the count condition subjects reported between 38 and 41 (mean = 40, SD = .6) of the 40 target stimuli demonstrating very good compliance with the task.
Overall activation in response to the task
All three conditions elicited activation in regions associated with a response to visual stimulation (occipital cortex) for both
Discussion
We investigated the effects of different response modalities on fMRI BOLD activation in a visual oddball paradigm, using passive, counting and responding variations of the task. As expected, we found visual cortex activation for target and frequent stimuli in all three conditions. Moreover, we observed target detection related activation in the count and respond conditions, but not in the passive condition. Motor related activation on the other hand was seen for the respond condition only. In
Conclusions
Our study clearly shows that the count and respond versions of the visual oddball task result in different patterns of BOLD activation that could both be attributed to ‘target detection’ if information on the respective other condition was not available. This methodological problem should be considered when interpreting existing own data or studies in the literature. This knowledge could also be used to design studies targeting specific aspects of the target detection network. The respond
Acknowledgments
We gratefully acknowledge participation of our subjects and would like to thank Frank Boers for technical assistance with stimulation paradigms and Jorge Arrubla for assistance with data collection.
Statement of interest
All authors have no disclosures/conflicts of interest to declare.
References (41)
- et al.
The gateway hypothesis of rostral prefrontal cortex (area 10) function
Trends Cogn. Sci.
(2007) - et al.
Assignment of functional activations to probabilistic cytoarchitectonic areas revisited
NeuroImage
(2007) - et al.
Single-trial discrimination for integrating simultaneous EEG and fMRI: identifying cortical areas contributing to trial-to-trial variability in the auditory oddball task
NeuroImage
(2009) - et al.
Intracerebral potentials to rare target and distractor auditory and visual stimuli. I. Superior temporal plane and parietal lobe
Electroencephalogr. Clin. Neurophysiol.
(1995) - et al.
Intracerebral potentials to rare target and distractor auditory and visual stimuli. II. Medial, lateral and posterior temporal lobe
Electroencephalogr. Clin. Neurophysiol.
(1995) - et al.
Generators of the late cognitive potentials in auditory and visual oddball tasks
Electroencephalogr. Clin. Neurophysiol.
(1998) - et al.
What is odd in the oddball task? Prefrontal cortex is activated by dynamic changes in response strategy
Neuropsychologia
(2004) - et al.
A global optimisation method for robust affine registration of brain images
Med. Image Anal.
(2001) - et al.
Improved optimization for the robust and accurate linear registration and motion correction of brain images
NeuroImage
(2002) - et al.
An adaptive reflexive processing model of neurocognitive function: supporting evidence from a large scale (n = 100) fMRI study of an auditory oddball task
NeuroImage
(2005)