Corsi Block-tapping Task (CBT) Performance Experiment

Abstract

The Corsi block-tapping task (CBT) is a widely used experimental tool for assessing visuo-spatial memory in both clinical and research contexts. However, whether information other than those spatial and visual (i.e., motor information) play also a role in CBT performance is still a matter of debate. Here, we investigated such issue through a crossed double dissociation design by observing how motor, visual, and spatial secondary tasks affect the performance on three versions of the CBT (standard, automatic and two-dimension). Results showed a double dissociation pattern, wherein two motor secondary tasks had larger effects when the CBT was administered by the examiner tapping on the blocks (standard version). A spatial secondary task had larger effects when the CBT was administered by automatically illuminating the blocks (automatic version). Finally, a visual secondary task had larger effects on a two-dimension, computerized version of the CBT. These findings suggest that memory for movements plays a relevant role in the CBT, and are especially relevant due to their implications for assessment of brain-damaged patients, besides providing further evidence of a fractionation of visuo-spatial memory into multiple sub-components.

Keywords: Corsi block-tapping task, visuo-spatial memory, memory for movements. Acknowledgements: The study was supported by a MIUR grant C26F014219 to F.F.

Introduction

The Corsi Block Tapping test (Milner, 1971; Corsi, 1972) has been widely used in cognitive psychology and in clinical neuropsychology to measure visuo-spatial memory (e.g., Kessels, de Haan, Kappelle, & Postma, 2003; Vandierendonck, Kemps, Fastame, & Szmalec, 2004) usually within the framework provided by the working memory model (Baddeley & Hitch, 1974). The standard apparatus consists of identical blocks irregularly arranged on a board. According to the standard administration procedure, but procedures vary widely among authors, the examiner taps on the blocks in randomized sequences of increasing length. The subject has to immediately reproduce each sequence, continuing until no longer accurate. Performance is measured as the longest sequence of blocks that is correctly reproduced.

Notwithstanding Baddeley (2001) reported the CBT as the task that is most closely related to the visuo-spatial short term memory, it is still not clear what of the two components, visual or spatial, it actually measures (Berch, Krikorian, & Huha, 1998; Quinn, 2008). This issue is relevant, since studies of both healthy individuals and brain-damaged patients demonstrated dissociable visual and spatial memory systems in humans (Klauer & Zhao, 2004; Carlesimo, Perri, Turriziani, Tomaiuolo, & Caltagirone, 2001). Such a fractionation of the visuo-spatial working memory is in fair agreement with evidence in primates of separate processing streams for visual and spatial features of objects (e.g., Goodale & Milner, 1992). Indeed, it has been proposed in both primates and humans that the dorsal visual system supports spatial working memory functions, and that the ventral visual system supports visual working memory for features of objects (e.g., Goldman-Rakic, 1987).

Evidences for a further fractionation of the visuo-spatial working memory were also reported, suggesting specific components of working memory for motor and kinesthetic information (Smyth, 1990). A close link between motor systems and visuo-spatial working memory was actually proposed since the very first studies about working memory (Baddeley, Grant, Wight, & Thomson, 1975). However, Smyth and her co-workers (Smyth & Pendleton, 1989) firstly suggested that a specific kinesthetic component of working memory might be responsible for the encoding and maintenance of remembered patterned movements (those aimed to bring the body parts into a specific configuration), whereas positional movements (movements targeted towards specific external spatial stimuli) appear to be encoded and maintained within the visuo-spatial sketchpad.

Notwithstanding the evidence favorable to a fractionation of the visuo-spatial working memory into multiple components, not necessarily independent one of each other, their relationship with the CBT has been actually scarcely investigated in literature. Though, the complex administration procedure of the CBT makes a more detailed analysis of the processes underlying the CBT strongly needed (Berch, Krikorian, & Huha, 1998). More interestingly, and maybe less obviously, the CBT might involve a memory for positional movements, because the administration procedure focuses on the movements of the examiner. However, the contribution of a memory for positional movements in the CBT task has never been investigated so far. It is also worth noting that computerized, two-dimension CBT versions have been frequently used (e.g., Vandierendonck, Kemps, Fastame, & Szmalec, 2004), albeit it is not known whether the standard and the computerized versions of the task are equivalent.

The present study aims at investigating the architecture of the visuo-spatial working memory as measured by the CBT, through a crossed double dissociation design (Dunn & Kirsner, 1988). We followed a standard dual-task procedure, using four secondary tasks aimed at interfering with the spatial, visual, and motor components of visuo-spatial working memory. They were crossed with three versions of the CBT: a) a standard version, wherein the sequences were given by the experimenter tapping on the blocks; in this version of the CBT the supposed motor/positional component was fully present; b) an “automatic” version, wherein the sequences were given by the blocks being illuminated; in this version the motor/positional component was removed from the task, while the spatial component was unaffected; c) a two-dimension version, presented on a computer monitor, wherein the sequences were given by the squares on the monitor changing their color; in this version, the spatial component of the task was reduced, albeit obviously not eliminated, by requiring the task to be performed on a 2D plane instead than in a 3D space.

Method

Participants. Forty-eight healthy, right handed individuals (mean age 22.4 years) participated in the experiment. All the participants reported normal or corrected-to-normal vision, and were naïve as to the purposes of the experiment.

Stimuli and apparatus. The apparatus was composed of eight translucent white 3 x 3 x 3 cm blocks, each one containing a red light emitting diode (LED). The blocks were fixed at random positions on a 23 x 30 cm translucent white board.

Procedure. Three administration procedures were used. In the standard procedure participants observed the experimenter tapping on the blocks, with his/her index finger, at a rate of one block per s, lifting the hand straight up before moving it to the next block (Standard). In the second procedure the to-be-remembered sequence was presented by the computer turning on and off the red LEDs inside the blocks, at a rate of one block per s (Automatic). A third, two-dimension version of the CBT was also used, as it is frequently used in literature as a substitute of the standard version. It was composed of eight blue squares appearing on the computer screen at the same relative positions as the 3D version described above. On each trial, the to-be-remembered sequence was indicated by the blocks changing color from blue to red and again to blue, at a rate of one block per s. The CBT was administered to all the participants according to the three procedures described above, in random order. Participants had to reproduce the sequence immediately after its administration, by tapping on the blocks using their index finger.

Sequences from 3 to 9 blocks in length were presented in ascending order, with two trials per length. All the fourteen sequences were administered to each participant. For each subject, different sequences, equated for paths’ length, were randomly assigned to the three versions of the test.

Each participant performed each version of the task both alone (single task condition), and along with one of four interference conditions (dual task condition), in random order: patterned-motor interference, motor interference, spatial interference and visual interference. In the patterned-motor interference condition, participants had to tap with their right index finger on the four corners of a mouse-pad, while the to-be-remembered sequence of blocks was administered. The movement had to be performed clockwise and continuously, at a rate of about one tap per s. Whereas this task is known to interfere with the CBT (Smyth & Pelky, 1992), it has both spatial and motor features that makes it difficult to disentangle their contribution. Thus, to remove the spatial component from this task we added a motor interference condition, wherein participants had to snap fingers with their right hand, while the to-be-remembered sequence of blocks was administered. The movement had to be performed continuously, in a regular manner (one snap per s, approximately). The experimenter controlled for the movement being correctly executed. In the spatial interference condition, participants were required to say aloud the side of each of a series of 1000 Hz tones randomly presented to their left or right ear through headphones, at 30 Db Spl with a constant inter-stimulus interval of 2 s. This listening task is supposed to interfere with the spatial component of the visuo-spatial sketchpad [18]. Finally, in the visual interference condition, one of three LEDs placed at the center of the board (one of three colored circles in the Two-Dimension Version) were turned on and off at a rate of one per s. On half the trials the regular sequence was violated, by turning on a differently colored led (on the 3D versions) or displaying a different colored circle (on the 2D version). At the end of each trial, participants were required to say whether a violation occurred on that trial.

Twelve participants were randomly assigned to the Patterned-motor, Motor, Spatial, and Visual Interference conditions, respectively.

The participants’ performance was measured as the longest sequence that was correctly reproduced at least once (memory span). Performance data were analyzed in a 3x2x4 ANOVA mixed design, with Version (standard, automatic, and two-dimension, within subjects), Condition (single task, dual task, within subjects), and Interference (patterned-motor, motor, spatial, and visual interference, between subjects) as factors.

Results

One participant in the Spatial Interference condition and two participants in the Visual Interference condition have been excluded from the following analyses because of the relatively large number of errors committed on the interference tasks. The remaining participants performed all the interference tasks at optimal levels, committing less than 3% of errors across visual and spatial interference tasks, and maintaining a regular mean rate of finger snapping and spatial tapping of about 1.2 per s. Figure 1 and Table 1 show the mean memory span length for each version of the CBT and for each interference condition.

A preliminary sphericity test failed to show any significant violation of the assumptions underlying the Version and the Version by Condition interference effects (p>.05 in all cases).

The analysis of performance data showed significant main effects of Condition (F_1,41=139.93, MSE=.42, p<.001), and Version (F_2,82=4.24, MSE=.63, p<.02). Also, the Interference by Version and the Condition by Version by Interference interactions were significant (F_6,82=3.61, MSE=.63, p<.01, F_6,82=4.33, MSE=.63, p<.01, respectively). The Version main effect was due to the performance at the two-dimension version of the CBT being slightly worse than at the other two versions of the task. While significant, though, such a difference was quite small (5.3 vs. 5.6), and will not be discussed any further. As main effect, participants performed better when they had to reproduce the target sequences they had seen in the single task condition than in the dual task condition, independently of the kind of interference and the version of the test (p<.05). Newman-Keuls testing also showed that the Interference by Version interaction was due to the individual’s performance on the three versions of the test being dependent on the type of interference that was administered. Again, such interaction is not relevant, since it concerns the average performance of individuals across the single and dual tasks conditions (i.e., without and with the interference). The Condition by Version by Interference interaction is more important with respect to the goal of investigating the processes underlying the performance at the CBT. Results of the Newman-Keuls test showed that in the single task condition there was no significant difference among the performances of the participants at the three versions of the CBT (p>.05 in all cases). This finding ensures that the administration procedure did not affect the difficulty of the task. However, the effects of the four kinds of interference upon the three versions of the CBT were very specific. Indeed, the patterned-motor and the motor interference tasks affected negatively the standard version of the test (p<.001 in both cases), but not the automatic and the two-dimension versions (p>.5 in both cases). The spatial interference task affected negatively the participants’ performance at the automatic version of the test (p<.001), but not the standard and the two–dimension versions (p>.05 in both cases). The visual interference task affected negatively the participant’s performance at the two-dimension version of the test (p<.001), but not the standard and the automatic versions (p>.05 in both cases). Importantly, such finding cannot be ascribed to the three interfering tasks being not equivalent with respect to each other, because of the triple dissociation procedure we employed.

Discussion

Results of the present experiment suggest that a component of working memory that deals with motor information has the major role in the standard version of the CBT. Indeed, the effects of both the motor and patterned-motor interference tasks were notably larger than those of the spatial and visual interference tasks in the standard version of the CBT. The crossed double dissociation general pattern of results strongly supports this interpretation. Indeed, the spatial interference task was more effective than both the motor interference tasks in the “automatic” version of the CBT, whereas only the visual interference task was effective in the two-dimension version of the CBT. Such result does not depend on confounding due to the three versions of the CBT being not equated in terms of difficulty, because in the single task condition the performance of the participants was the same in the three versions of the test. Also, it does not depend on the spatial interference task involving a verbal coding of the spatial locations where the tones came from, as the phonological loop has been shown to be not involved in the CBT (e.g., Vandierendonck, Kemps, Fastame, & Szmalec, 2004).

The finding that the performance on the standard version of the CBT largely depends on individuals coding the movements of the examiner is in fair agreement with the hypothesis that a component of working memory that deals with motor information actually exists, and is independent of the component of working memory that deals with spatial information (e.g., Smyth & Pendleton, 1990). It is also in fair agreement with the growing body of neurophysiological and psychological studies that suggest a close link between observing and performing an action (e.g., Rizzolatti, Fadiga, Gallese, & Fogassi, 1996). Interestingly, van Asselen and coworkers (van Asselen, Kessels, Sebastiaan, Neggers, Kappelle, Frijns, et al. 2006) have recently interpreted results of a study on stroke patients as suggesting that the right dorsolateral prefrontal cortex (DLPFC) and the right posterior parietal cortex (PPC) are involved in keeping spatial information in memory over a short time period, as was assessed with the CBT. While the involvement of both the DLPFC and the PPC in spatial memory tasks is not new (e.g., Walter, Bretschneider, Groen, Zurowski, Wunderlich, Tomczak, et al. 2003), it is worth noting that this is not at variance with the hypothesis that a specific component of working memory for positional movements is involved in the CBT. For instance, lesion and physiological studies have shown that the DLPFC has a crucial role in visuospatial control of actions and visuomotor transformations (e.g., Curtis & D’Esposito, 2004). Indeed, Hoshi (Hoshi, 2006) in a recent review suggested that the dorsal part of the DLPFC is involved in representing processed motor information, such as arm use or target location, and in integrating multiple classes of information for planning action. Similarly, the PPC is involved in visuomotor transformation, and is thought to serve as a sensorymotor interface for visually guided eye and limb movements (Buneo & Andersen, 2006). Moreover, evidence has been recently provided that, within the fronto-parietal network of brain regions involved in learning spatial sequences, two partially segregated neural systems are involved in processing spatial sequences in reaching and navigational space (Nemmi, Boccia, Piccardi, Galati & Guariglia, 2013), supporting the idea of a further fractionation of visuo-spatial memory into multiple sub-components. Though, more research is needed in order to specify the relationship between the complex functional architecture of the DLPFC – PPC system and the specific features of the working memory components, including those measured by the CBT.

Finally, it is worth noting that the motor and spatial interference tasks affected only marginally the performance on the two-dimension version of the CBT. Such a result suggests that the two-dimension and the standard versions of the CBT cannot be considered as equivalent. This finding is especially relevant because recently two-dimensions, computerized versions of the CBT have been used rather frequently in clinical and experimental settings (Vandierendonck, Kemps, Fastame, & Szmalec, 2004; Joyce, & Robbins, 1991).

In conclusion, the present study shows that the performance on the Corsi block-tapping task depends largely on a component of working memory specifically dealing with motor information and that this component is independent of that component of working memory that deals with spatial information.

Beside providing further evidence of a fractionation of visuo-spatial memory into multiple sub-components, present findings have important implications for clinical assessment of brain-damaged patients and should be taken into account when interpreting the performance on the CBT for neuropsychological rehabilitation treatments in clinical settings.