Functional magnetic resonance imaging of mental rotation and memory scanning: a multidimensional scaling analysis of brain activation patterns1

https://doi.org/10.1016/S0165-0173(97)00060-XGet rights and content

Introduction

Mental rotation 6, 18, 19and memory scanning 4, 12, 20are typical examples of cognitive operations presumably involved in various tasks [7]. The original tasks involved judgements to be indicated by key presses 18, 19or verbal responses [20], whereas recent variants 4, 6, 12required directed movements as responses. The cardinal sign of both mental rotation and memory scanning tasks is the increase of the response time with task demands, namely the angle of rotation 6, 18, 19or of the number of items in the list scanned 4, 12, 20. The rates of processing information in these two kinds of tasks are uncorrelated [12], which suggests that different brain mechanisms may be involved. In contrast, the rates of rotation of a figure or a movement direction are positively correlated [12], which suggests that common aspects of brain mechanisms may be involved in widely different cases of mental rotation.

The overall brain mechanisms underlying mental rotation and memory scanning are largely unknown. With respect to mental rotation, a consistent finding of electrophysiological and other studies has been that parietal areas are involved in mental rotation of visual images 1, 2, 3, 10, 11, 14, 15, 16, 21, 22, 23, 25, 26. With respect to relations between brain activation and task performance, recent studies using functional magnetic resonance imaging (fMRI) 22, 23have indicated an involvement of the parietal areas, bilaterally, with performance outcome and an association of the precentral gyrus of the right hemisphere with the rate of mental rotation in the Shepard and Metzler [19], 3D image paradigm [23].

Neurophysiological studies in the motor cortex of behaving monkeys have provided an insight into the neural processes involved in the mental rotation 5, 9and memory scanning tasks [13]that required directed movements as indicators of response; namely, a slow rotation of a directional population signal in mental rotation 5, 9and an abrupt switch in direction of that signal in memory scanning [13]. Although these studies addressed the question of neural processing within a particular cortical area (i.e. the motor cortex), several brain areas are probably involved in the performance of these tasks. It is also obvious that tasks can resemble (or differ from) each other along different dimensions. For example, some tasks involve mental rotation whereas others involve memory scanning; and some involve mental operations on visual images whereas others involve operations on movement representations. The constellation of brain areas that are differentially involved in, and essentially represent, those dimensions can be revealed best by techniques that allow the determination of engagement of a number of areas by a given task. This information can be obtained accurately in single subjects by fMRI which we employed in this study.

Typically, several areas are activated during performance of a given task, and this set of areas frequently overlaps with the set of another task. This, in turn, implies that task-related information is processed in a distributed fashion by various areas. In the present study, we sought to recover this information from binary functional activation maps using weighted multidimensional scaling.

Section snippets

Tasks

Ten healthy subjects (5 women and 5 men) were paid volunteers and performed the following tasks with the approval of the Institutional Review Board of the University of Minnesota. All subjects were right handed and performed with the right hand. The experimental tasks used, and the corresponding control tasks, are illustrated in Fig. 1. Each subject performed all of the following tasks in a random order.

In the visual mental rotation task (task A in Fig. 1), the letter G was shown rotated from

Results and discussion

As mentioned above, the cardinal performance feature of the tasks employed is the linear increase of the response time with (i) the angle of rotation, in the mental rotation tasks (A and B in Fig. 1), and (ii) the number of list stimuli, in the memory scanning tasks (C and D in Fig. 1). The slopes of these equations are characteristic of the respective processes. All subjects in the present study showed these relations. Data and fitted regression lines are illustrated for one subject in Fig. 1,

Conclusions

Although functional brain imaging provides a powerful tool by which to identify brain areas involved in performance of particular cognitive tasks, the resulting complex functional activation maps are difficult to interpret and even more difficult to relate, as a whole pattern, to the richness of the tasks used. In this paper, we demonstrated the successful application of weighted multidimensional scaling to as complicated brain activation patterns as those resulting from the study of 30 brain

Acknowledgements

This study was supported by United States Public Health Service grants NS32919 and RR088079, the Unites States Department of Veterans Affairs and the American Legion Chair in Brain Sciences.

First page preview

First page preview
Click to open first page preview

References (27)

  • A.P. Georgopoulos et al.

    Mental rotation of the neuronal population vector

    Science

    (1989)
  • A.P. Georgopoulos et al.

    Cognitive spatial-motor processes. 1. The making of movements at various angles from a stimulus direction

    Exp. Brain Res.

    (1987)
  • S.-G. Kim et al.

    Potential pitfalls of functional MRI using conventional gradient-recalled echo techniques

    NMR in Biomedicine

    (1994)
  • Cited by (78)

    • Viewer and object mental rotation in young adults with psychotic disorders

      2022, Schizophrenia Research
      Citation Excerpt :

      However, nowadays it is recognized that spatial cognition includes many different functions (e.g., spatial attention, navigation, spatial imagery, etc.) which some were found to be lateralized and others not (Cona and Scarpazza, 2019). In regards to mental rotation, neuroimaging studies have found that mental rotation was associated with the activation of a bilateral frontoparietal network (Cona and Scarpazza, 2019; Tagaris et al., 1998). Interestingly, the activity of that bilateral network increased with the angle of object mental rotation but decreased with the angle of subject rotation (Carpenter et al., 1999; Gogos et al., 2010; Keehner et al., 2006; Milivojevic et al., 2009), which is evidence that the angle coding depends on the reference frame of the mental rotation.

    • Differential involvement of left and right frontoparietal areas in visuospatial planning: An rTMS study

      2020, Neuropsychologia
      Citation Excerpt :

      This pattern of results showed that, more than gender, individuals' performance level could account for different neuronal activation patterns during higher-level cognition. Other studies have already shown that variations in brain activation are more strongly related to performance than to gender (Tagaris et al., 1998; Unterrainer et al., 2000) while still observing better performance in males than in females at the behavioural level. On the other hand, Boghi et al. (2006) showed, in an fMRI recording of TOL tasks at different difficulty levels, that there were different patterns of activation and concluded that gender-specific strategies could exist, with males relying more on visuospatial abilities and females relying more on executive processing.

    • Brain Correlates of Human Leukocyte Antigen (HLA) Protection in Gulf War Illness (GWI)

      2016, EBioMedicine
      Citation Excerpt :

      We sought to derive such an arrangement by applying MDS, a venerable method for uncovering basic dimensions in a multidimensional dataset (Shepard, 1980, 1988). The MDS is a well-established dimension-reduction method (Borg and Groenen, 2010) that has proved valuable in psychological (Shepard, 1980, 1988; Whang et al., 1999), psychiatric (Stephane et al., 2003), and neuroscientific applications (Leuthold et al., 2013; Tagaris et al., 1998; Tzagarakis et al., 2009; Young, 1992; Young and Yamane, 1992; Young et al., 1995). It typically reveals underlying dimensions along which variables are ordered, dimensions that may be hidden when embedded in a multidimensional space.

    • A key role for experimental task performance: Effects of math talent, gender and performance on the neural correlates of mental rotation

      2012, Brain and Cognition
      Citation Excerpt :

      As previously mentioned, this region was strongly activated by the task. Using the standard paradigm, a bilateral activation of SPL (instead of IPL) was obtained as a positive neural correlate of mental rotation accuracy (Tagaris et al., 1996a, 1996b, 1997, 1998). As predicted, the effect of the experimental task performance factor (bilateral IPL) showed a substantial overlap with the effects of math talent (left IPL/SPL) and gender (left IPL), so that the left IPL could be characterized as the convergent zone for capability and performance related effects on mental rotation related brain activations.

    • Pattern-information analysis: From stimulus decoding to computational-model testing

      2011, NeuroImage
      Citation Excerpt :

      While block-designs and conventional event-related designs typically assume a stimulus grouping, ungrouped-events designs (Kriegeskorte et al., 2008b; see also Aguirre 2007) avoid this assumption. An ungrouped-events design enables us to discover stimulus dimensions represented in a given region, for example using data-driven multivariate techniques like multidimensional scaling (for applications to fMRI data see Edelman et al., 1998; Tagaris et al., 1998; O'Toole et al., 2005; Kriegeskorte et al., 2008a, 2008b). Conversely, we can look for regions representing a given stimulus dimension (or conforming to a given computational model), for example with a searchlight approach (Kriegeskorte et al., 2006).

    View all citing articles on Scopus
    1

    Published on the World Wide Web on 24 February 1998.

    2

    Current address: Department of Neurology, General Athens Hospital, Athens, Greece.

    View full text