Comparison of baseline conditions to investigate syntactic production using functional magnetic resonance imaging
Introduction
Results of neuroimaging studies do not reflect single processes, but instead reflect contrasts between cognitive processes in two conditions or tasks (Newman et al., 2001). The ability to elucidate neural correlates of cognitive functions using functional magnetic resonance imaging (fMRI) is affected by many factors including the selection of an appropriate baseline condition. Discrepancies between functional imaging studies of cognition can often be explained by comparisons of similar target tasks to different control tasks Damasio et al., 1996, Ishai et al., 1999. A resting state in which the participant is instructed not to engage in specific cognitive processes is commonly used as a baseline condition in fMRI paradigms. However, fundamental neural activity that exists even in rest conditions makes using rest as a control or baseline state in fMRI problematic (Gusnard and Raichle, 2001), because activity during a resting baseline may vary unpredictably. Furthermore, it has been shown that this ongoing activity during rest may involve the same brain regions activated during the target task (Binder et al., 1999). A number of neuroimaging studies have even reported that activity in some regions during rest is greater than in some cognitive tasks Schulman and Fiez, 1997, Stark and Squire, 2001, Raichle et al., 2001.
The use of perceptual tasks, defined as activities utilizing external information, during control states is thus desirable to better control the baseline neural activity that is possible during resting states characterized by variable and unpredictable activity. However, experimental designs that use a cognitive subtraction paradigm have been criticized in part due to the potential problems in the logic of the “pure insertion” assumption that this technique makes Friston et al., 1996, Jennings et al., 1997, Sidtis et al., 1999. Cognitive subtraction methods hold that the neural processes underlying behavior are additive and assume it is possible to insert or delete cognitive components to a task without affecting other preexisting processes (Friston et al., 1996). Such designs typically employ hierarchical baseline conditions, and comparisons are made between tasks that have overlapping but seemingly separable cognitive components. Presumably, this allows the investigator to parse apart the process of interest. However, critics of this methodology claim that it fails to account for the interaction of the cognitive components of a task (Friston et al., 1996). In fact, several studies have provided evidence against task additivity and suggest that processes combine in a nonadditive or interactive fashion Friston et al., 1996, Jennings et al., 1997, Sidtis et al., 1999. Despite the criticism directed toward the logic of “pure insertion”, not all cognitive subtraction paradigms presume pure insertion. Newman et al. (2001) asserted that the choice of baseline condition differentially affects observed activation based on the notion of task overlap; different activation patterns emerged from the same experimental condition when measured with different baselines. This could reflect either “impure insertion” or different task overlaps with purely linear, independent, noninteractive processes. The complex nature of task comparisons in functional neuroimaging studies highlights the importance of identifying appropriate baseline and control tasks to adequately address the experimental questions and correctly interpret the data to discern the neural correlates of cognitive activity.
The purpose of the current study was to investigate the most appropriate baseline task, picture naming or passive viewing of nonsense objects, to isolate syntactic and related processes with silent sentence generation during fMRI. Our goal in acquisition of this knowledge is direct application of fMRI in a post-stroke rehabilitation setting to reveal neural correlates of rehabilitative change. The majority of functional neuroimaging studies are designed to compare task with experimental control activity to isolate cognitive processes using group analyses. However, single-subject data analyses are necessary for conditions such as stroke, where lesion boundaries vary considerably between subjects, and averaging images across subjects would result in a gross underestimation of perilesional activity. The aim of the current study was to determine the baseline condition that maximizes sensitivity in detecting signal change related to the target task, simple sentence generation, while still achieving specificity in isolating the activity associated with that cognitive task in both group and individual-subject level analyses to inform longitudinal single-case studies of stroke rehabilitation.
Chaterjee and Maher (2000) pointed out that the neural substrates of sentence processing have not been fully explained. Based on the lesion literature, posterior frontal regions of the left hemisphere (e.g., Brodmann's areas 44, 45, and 6) have been suspected of involvement in syntactic processing (Nadeau, 1988), including comprehension and production. Given the involvement of left prefrontal, Rolandic, insular, and anterior parietal cortices in chronic Broca's aphasia, where impaired sentence production is an identifying characteristic of the syndrome (Alexander, 2003), it seems likely that sentence processing engages a broader network than the posterior frontal cortex. Indeed, Nadeau (1988) has speculated that posterior perisylvian cortices (i.e., Brodmann's areas 21, 22, 39, and 40) in the left hemisphere may be important for processing morphological aspects of grammar during sentence production. Certainly, these posterior perisylvian regions appear to be involved in sentence comprehension Caplan et al., 1996, Selnes et al., 1983, Tramo et al., 1988.
Recent functional neuroimaging studies of syntactic processing in healthy adults have implicated the left inferior frontal cortex, particularly Broca's area Grodzinsky, 2000, Kuperberg et al., 2000, Zurif et al., 1993. A variety of syntactic functions have been attributed to Broca's area, including comprehending lexical-semantic information and syntactic structure (Dapretto and Bookheimer, 1999), processing syntactic ambiguities (Stowe et al., 1998), detecting syntactic anomalies and errors (Ni et al., 2000), and processing syntactically more complex sentences Caplan, 2001, Caplan et al., 2000, Just et al., 1996. Other cortical regions involved in syntactic processing include the posterior portion of the superior temporal gyrus (Just et al., 1996), the angular gyrus and supramarginal gyrus (Caplan, 2001), and the superior parietal lobe Carpenter et al., 1999, Just et al., 1996, Newman et al., 2001. However, all of these neuroimaging studies have relied on sentence comprehension (using auditory or orthographic presentation) to study syntactic processing. Few functional imaging studies, if any, have investigated sentence production in healthy adults, and we are unable to identify such studies that have examined sentence production in patients with chronic nonfluent aphasia. It is possible that syntactic generation may involve regions within the left hemisphere perisylvian syntax network differentially than syntactic comprehension.
Finally, given the role that the literature has ascribed to Broca's area in syntactic processing, including syntactic generation, we concluded that the most appropriate baseline comparison would demonstrate activation in this area. In particular, for a paradigm that will be used in stroke rehabilitation, it is desirable that activity in Broca's area be detectable at the individual subject level.
Section snippets
Participants
Ten right-handed neurologically normal adults (three men, seven women) between the ages of 20 and 58 years (mean age = 40.5 years, SD = 15.0) with education between 13 and 20 years (mean education = 16.4 years, SD = 2.2) participated in the current study. All participants were native English speakers and were right-handed. Volunteers were excluded if they had a history of a psychiatric disorder, substance abuse, or were currently taking psychotropic medication. Informed consent was obtained
Individual-subject analysis
Because of the importance of Broca's area in syntactic processing, an analysis was conducted to determine whether use of one baseline task or the other offered an advantage in imaging Broca's area activity at the individual-subject level. Clusters of significant activity were found in Broca's area for 8 of 10 subjects in at least one of the tasks. After square-root transformation to reduce the extreme positive skew of the data, it was found that the use of nonsense object viewing as the
Discussion
We empirically demonstrated differences in activation patterns during sentence production based on two different baseline tasks (picture naming and nonsense object viewing). At both the individual and group level, results of the current study suggest that passive viewing of nonsense objects serves as a more sensitive baseline comparison than picture naming for investigation of sentence generation. Given our analysis of regions involved in syntactic processing, more confidence can be placed in
Acknowledgements
This research was supported by NIH grant P50-DC03888 and is based upon work supported in part by the Office of Research and Development, RR&D Service, Department of Veterans Affairs.
References (37)
- et al.
Overt verbal responding during fMRI scanning: empirical investigations of problems and potential solutions
NeuroImage
(1999) - et al.
Time course of fMRI-activation in language and spatial networks during sentence comprehension
NeuroImage
(1999) AFNI: software for analysis and visualization of functional magnetic resonance neuroimages
Comput. Biomed. Res.
(1996)- et al.
Form and content: dissociating syntax and semantics in sentence comprehension
Neuron
(1999) - et al.
The trouble with cognitive subtraction
NeuroImage
(1996) - et al.
Cognitive subtractions may not add up: the interaction between semantic processing and response mode
NeuroImage
(1997) - et al.
Recognizing words, pictures, and concepts: a comparison of lexical, object, and reality decisions
J. Verbal Learn. Verbal Behav.
(1984) - et al.
Are brain functions really additive?
NeuroImage
(1999) - et al.
Empirical analysis of BOLD fMRI statistics: I. Spatially unsmoothed data collected under null-hypothesis conditions
NeuroImage
(1997) - et al.
An on-line analysis of syntactic processing in Broca's and Wernicke's aphasia
Brain Lang.
(1993)
Aphasia: clinical and anatomic issues
Anterior cingulate and the monitoring of response conflict: evidence from an fMRI study of overt verb generation
J. Cogn. Neurosci.
Conceptual processing during the conscious resting state: a functional MRI study
J. Cogn. Neurosci.
Event-related fMRI tasks involving brief motion
Hum. Brain Mapp.
Functional neuroimaging studies of syntactic processing
J. Psycholinguist. Res.
Location of lesions in stroke patients with deficits in syntactic processing in sentence comprehension
Brain
Activation of Broca's area by syntactic processing under conditions of concurrent articulation
Hum. Brain Mapp.
Grammar and agrammatism
Cited by (19)
Cerebellum and Grammar Processing
2016, The Linguistic CerebellumNon-invasive mapping of bilateral motor speech areas using navigated transcranial magnetic stimulation and functional magnetic resonance imaging
2015, Journal of Neuroscience MethodsCitation Excerpt :When generating words from visually presented letter, the baseline task has been often a presentation of an asterisk/fixation cross/other symbol for visual activation (e.g. (Besseling et al., 2013; Black et al., 2014; Somers et al., 2011; Zaca et al., 2013)). However, fundamental neural activity that exists even in rest conditions makes using rest as a control or baseline state in fMRI problematic (Gusnard et al., 2001), because activity during a resting baseline may vary unpredictably (Peck et al., 2004). In this respect, a simple but standardized baseline task was chosen.
Syntactic priming and the lexical boost effect during sentence production and sentence comprehension: An fMRI study
2013, Brain and LanguageCitation Excerpt :Far fewer neuroimaging studies have looked at sentence production. They show that left inferior frontal regions (Haller, Radue, Erb, Grodd, & Kircher, 2005; Indefrey, Hellwig, Herzog, Seitz, & Hagoort, 2004; Indefrey et al., 2001; Peck et al., 2004) and sometimes also left posterior temporal regions (Menenti, Segaert, & Hagoort, 2012) are implicated in syntactic processing in production. In our own recent work aiming to compare the brain regions involved in comprehension and production, we found that syntactic processing in the two modalities is subserved by largely the same brain regions—the left IFG and left posterior MTG (Menenti, Gierhan, Segaert, & Hagoort, 2011; Segaert, Menenti, Weber, Petersson, & Hagoort, 2012).
Imaging of brain tumors: Functional magnetic resonance imaging and diffusion tensor imaging
2010, Neuroimaging Clinics of North AmericaCitation Excerpt :Depending on the target location, a nonstimulating task such as fixation on a cross or an opposing task can be chosen. For example, a baseline task of passively viewing nonsense objects has been shown to improve detection of signal in Broca's area with sentence generation tasks.21 Direct electrical cortical stimulation (ECS) is considered the gold standard for functional mapping.
Laterality index in functional MRI: methodological issues
2008, Magnetic Resonance ImagingCitation Excerpt :For instance, when a phoneme discrimination task was compared to three baseline conditions (rest, tone monitoring and passive listening), the baseline conditions systematically affected the amount of activation in different language areas [78], demonstrating different LI values with different baseline conditions. In addition, a recent study has investigated the most appropriate baseline task (picture naming, or passive viewing of nonsense objects) to isolate syntactic processes [79]. This showed that activity in language areas (e.g., Broca's area) was dependant on the baseline used, with more activation in these areas when the baseline was passive viewing of nonsense objects than picture naming.
- 1
Current address: Department of Medical Physics and Radiology, Memorial Sloan-Kettering Cancer Center, New York, NY 10021, USA.