Elsevier

NeuroImage

Volume 18, Issue 2, February 2003, Pages 367-374
NeuroImage

Regular article
Frontostriatal system in planning complexity: a parametric functional magnetic resonance version of tower of london task

https://doi.org/10.1016/S1053-8119(02)00010-1Get rights and content

Abstract

In the present study, we sought to investigate which brain structures are recruited in planning tasks of increasing complexity. For this purpose, a parametric self-paced pseudo-randomized event-related functional MRI version of the Tower of London task was designed. We tested 22 healthy subjects, enabling assessment of imaging results at a second (random effects) level of analysis. Compared with baseline, planning activity was correlated with increased blood oxygenation level-dependent (BOLD) signal in the dorsolateral prefrontal cortex, striatum, premotor cortex, supplementary motor area, and visuospatial system (precuneus and inferior parietal cortex). Task load was associated with increased activity in these same regions. In addition, increasing task complexity was correlated with activity in the left anterior prefrontal cortex, a region supposed to be specifically involved in third-order higher cognitive functioning.

Introduction

Planning, i.e., the ability to achieve a goal through a series of intermediate steps, is an essential component of higher-order cognitive processes, such as problem solving. Central to the concept of the neural substrate for cognitive planning is the position of the dorsolateral prefrontal cortex, but other brain regions, like the premotor, cingulate, and insular cortices and the striatum, have also been implicated in cognitive planning Owen 1997, Robbins 1998. Whereas until recently insight into the role of brain structures in planning processes was based primarily on lesion studies in patients, contemporary studies are able to employ human brain imaging techniques. The results of such functional neuroimaging studies have suggested that a distributed corticocortical and corticosubcortical network is being activated during complex planning tasks, although there is yet no complete agreement about the contribution of the various components of such neuronal networks.

A direct and very frequently used test to probe planning processes is the Tower of London task, originally developed by Shallice (1982). The Tower of London task has been modified in various ways to make it suitable for testing different components of cognitive planning in clinical and experimental psychological settings. The complexity of the task can range from trivial, requiring only one obvious move, to highly complex. The more complex versions of the task require the planning of intermediate, in many instances counterintuitive, steps to successfully solve a problem. Individuals with decreased prefrontal functioning, such as those with neurosurgical lesions (Shallice, 1982), neurodegenerative diseases like Huntington’s and Parkinson’s diseases Dagher et al 1999, Watkins et al 2000, Hodgson et al 2002, or psychiatric disorders like obsessive–compulsive disorder (Veale et al., 1996) and schizophrenia (Pantelis et al., 1997), are impaired on the task, in particular when complexity increases.

The Tower of London task is also very suitable for studying in healthy subjects the patterns of brain activation in the context of planning tasks using positron emission tomography (PET) Owen et al 1996, Baker et al 1996, Dagher et al 1999, Rowe et al 2001 or single-photon emission computed tomography (SPECT) Rezai et al 1993, Morris et al 1993. The results of these imaging studies agree on the involvement of the dorsolateral prefrontal cortex and parieto-occipital regions (visuospatial system) during planning. However, activation of other brain regions, such as the cingulate and insular cortices and the striatum, has not been found across all mentioned studies. Furthermore, the results of a functional MRI study using the Tower of London task of Lazeron et al. (2000) confirmed activation of the dorsolateral prefrontal, parietal, cingulate, and insular cortices but failed to show striatal activation. Several methodological differences between these studies may explain the inconsistencies in the results. First, different baseline conditions have been used, ranging from low-level conditions (watching a blank screen) to conditions involving motor activity (moving beads without a goal) or cognitive functioning (counting beads). Second, differences in the execution of the task exist: whereas in some studies subjects used a touch screen to perform the task Owen et al 1996, Dagher et al 1999, Rowe et al 2001, in other designs subjects were requested to execute their moves mentally Baker et al 1996, Lazeron et al 2000. A third likely source of inconsistency concerns the differences in task load. Because in subtraction designs the specificity of active versus baseline differences for the (cognitive) function under study is often questionable Friston et al 1996, Sidtis et al 1999, several researchers have used designs in which task load during performance of the Tower of London task was manipulated. For example, Baker et al., (1996), in comparing “difficult” (four or five moves) and “easy” trials (two or three moves), found task load to be correlated with increased regional cerebral blood flow (rCBF) in rostrolateral and dorsolateral prefrontal cortex, and the parietal and cingulate cortices. Owen et al. (1996) reported increased activation in striatum and thalamus, but not in cortical areas. Lazeron et al. (2000) did not find any difference between their “easy” and “difficult” versions, but they used a more difficult design comparing two to four moves with five to seven moves. The only study employing a parametric approach in studying task load (Dagher et al., 1999) used five levels of planning and showed a correlation between planning complexity and increased rCBF in the right caudate nucleus and prefrontal and cingulate areas. The involvement of visuospatial areas in the parietal lobe, however, seemed not to be related to the task load. A general methodological drawback in the studies performed so far is the lack of control over performance differences, due to the use of blocked designs. Since performance is likely to deteriorate at higher complexity, this is especially relevant for parametric tasks.

It must be assumed that the various cortical and subcortical brain structures that have been demonstrated to be involved in planning processes execute these functions as “nodes” in distributed neuronal networks (Owen, 1997). Employing a neuronal network model for planning processes, Dehaene and Changeux (1997) proposed multiple hierarchic levels with ascending and descending streams for execution of planning and evaluation of this behavior, respectively. This model also stresses that planning behavior cannot be related to a single region, but rather relies on multiple neuronal circuits coding for specialized subprocesses such as working memory, plan generation, and internal reward. These subprocesses may differentially contribute to different levels of planning behavior; some subprocesses may be relatively independent of task load, while other subprocesses are involved mainly at higher levels of planning behavior. The main aspect in increasing planning complexity is the increasing amount of subgoals and counterintuitive moves, while holding in mind the overall main goal. Koechlin et al. (2000) introduced the term branching for this process of integrating working memory with attentional resource allocation, a third-order higher cognitive function. They found, using a multitask experiment, branching to rely on activation of the rostral part of the prefrontal cortex. To be able to investigate these specific subprocesses and to differentiate between different brain regions in their contribution to different levels of planning behavior, parameterization of the task is essential. Therefore, the main aim of the present study was to determine the effects of task load in planning execution.

The present study employs event-related functional MRI as a method to study activation patterns on a trial-by-trial basis. This enabled us (i) to randomize items with respect to task complexity, and (ii) to include correct responses only. A large number of subjects (n = 22) were included, with the aim of performing second-level or random effects analyses, allowing for generalization of our results. On the basis of the considerations above, we hypothesized that increased planning complexity would be associated with increased activity in dorsolateral prefrontal, visuospatial, and striatal regions. In addition, we expected rostral areas of the prefrontal cortex to be specially involved in the higher levels of planning execution.

Section snippets

Subjects

Twenty-two healthy right-handed subjects (11 men and 11 women; mean age 29.9 years, range 21–49 years) performed the Tower of London task, while functional MRI data were collected. Subjects were recruited among university students and staff. The ethical review board of the VU Medical Centre approved of the study and all participants provided written informed consent.

Task paradigm

To ensure that participants were familiar with the procedure, the test was explained and practiced outside the scanner before MR

Task performance

Mean performance scores were 94.2% (SD = 1.68) for baseline (mean RT = 3.7 s, SD = 0.88), 97.6% (SD = 2.02) for one move (mean RT = 4.4 s, SD = 1.11), 95.4% (SD = 3.99) for two moves (mean RT = 5.8 s, SD = 1.27), 96.7% (SD = 3.58) for three moves (mean RT = 7.4 s, SD = 1.90), 89.9% (SD = 7.17) for four moves (mean RT = 10.1 s, SD = 2.51), and 82.2% (SD = 12.59) for five moves (mean RT = 15.0 s, SD = 4.02). Both the decrease in performance (F(4,18) = 12.4, P < 0.0001) and the increase in RT (F

Discussion

In this article we present a parametric event-related functional MRI version of the Tower of London task, suitable for investigating the effects of increasing complexity in planning processes. In agreement with the results of earlier imaging studies, the present fMRI data show that the process of planning, compared with baseline, was strongly correlated with activation of the right dorsolateral prefrontal cortex (BA 9 and BA 46), bilateral premotor cortex (BA 6 and BA 8), bilateral precuneus

Acknowledgements

This work was supported by the Dutch Organization for Scientific Research (NWO), MW 940-37-018.

References (26)

  • L.H. Watkins et al.

    Impaired planning but intact decision making in early Huntington’s diseaseimplications for specific frontostriatal pathology

    Neuropsychologia

    (2000)
  • G.E. Alexander et al.

    Parallel organization of functionally segregated circuits linking basal ganglia and cortex

    Annu. Rev. Neurosci.

    (1986)
  • A. Baddeley

    The fractionation of working memory

    Proc. Natl. Acad. Sci. USA

    (1996)
  • Cited by (201)

    View all citing articles on Scopus
    View full text