Differentiated parietal connectivity of frontal regions for "what" and "where" memory

Abstract

In a previous meta-analysis across almost 200 neuroimaging experiments, working memory for object location showed significantly stronger convergence on the posterior superior frontal gyrus, whereas working memory for identity showed stronger convergence on the posterior inferior frontal gyrus (dorsal to, but overlapping with Brodmann's area BA 44). As similar locations have been discussed as part of a dorsal frontal-superior parietal reach system and an inferior frontal grasp system, the aim of the present study was to test whether the regions of working-memory related "what" and "where" processing show a similar distinction in parietal connectivity. The regions that were found in the previous meta-analysis were used as seeds for functional connectivity analyses using task-based meta-analytic connectivity modelling and task-independent resting state correlations. While the ventral seed showed significantly stronger connectivity with the bilateral intraparietal sulcus (IPS), the dorsal seed showed stronger connectivity with the bilateral posterior inferior parietal and the medial superior parietal lobule. The observed connections of regions involved in memory for object location and identity thus clearly demonstrate a distinction into separate pathways that resemble the parietal connectivity patterns of the dorsal and ventral premotor cortex in non-human primates and humans. It may hence be speculated that memory for a particular location and reaching towards it as well as object memory and finger positioning for manipulation may rely on shared neural systems. Moreover, the ensuing regions, in turn, featured differential connectivity with the bilateral ventral and dorsal extrastriate cortex, suggesting largely segregated bilateral connectivity pathways from the dorsal visual cortex via the superior and inferior parietal lobules to the dorsal posterior frontal cortex and from the ventral visual cortex via the IPS to the ventral posterior frontal cortex that may underlie action and cognition.

Fig. 1

a Segregation of the frontal cortex as revealed by a coordinate-based meta-analysis of working memory studies (Rottschy et al. 2012). Regions where experiments on object location showed a significantly higher convergence of reported activations than those probing memory for object identity are shown in red. Regions showing stronger convergence of activation in experiments on object identity are displayed in green. The left ventral region serves as exemplary seed to illustrate the meta-analytic connectivity modelling (MACM) approach in b–d. b Activation foci of all experiments in the BrainMap Database, which show at least one activation in the seed region (left posterior inferior frontal gyrus). c The reported coordinates, which are shown in b are treated as probability distributions, which indicate that the “true” locations are modelled as 3D Gaussians. d Random effect inference against a null-distribution of random spatial association across studies

Fig. 2

Time series of the dorsal seed region (posterior superior frontal gyrus) in a single subject is shown in grey. From the same subject, the time series of an uncorrelated voxel is shown in red and that of a correlated voxel in green

Fig. 3

a Regions showing significantly stronger task-based (MACM) functional connectivity with the ventral as opposed to the dorsal posterior frontal seed. b Regions showing significantly stronger task-based (MACM) functional connectivity with the dorsal as opposed to the ventral posterior frontal seed

Fig. 4

a Brain regions showing significantly stronger task independent (resting state) connectivity with the ventral as opposed to the dorsal frontal seed. Please note, that the parietal activation was predominantly located in the IPS but projects to the IPL in the lateral view. b Brain regions showing significantly stronger task independent (resting state) connectivity with the dorsal as opposed to the ventral seed region

Fig. 5

a Conjunction across task dependent (MACM) and task independent contrast analyses. Regions, which showed stronger connectivity with the posterior inferior frontal cortex are shown in green, while those regions, which showed stronger connectivity with the posterior superior frontal cortex are shown in red. b Conjunction across both approaches [task dependent (MACM) and independent (resting state) functional connectivity] and seeds (ventral and dorsal posterior inferior and posterior superior frontal cortex)

Fig. 6

a Conjunction across task dependent (MACM) and task independent (resting state) connectivity differences between parietal regions, which showed connectivity with the dorsal and ventral frontal seed in the above shown analysis (cf. Fig. 5a). Regions, which showed stronger connectivity with the IPS are shown in green, while those regions, which showed stronger connectivity with the SPL/IPL are shown in red. b Significant differences in connectivity between the IPS and SPL/IPL in the posterior frontal cortex. Here it is shown, that IPS showed stronger connectivity with the bilateral ventral seeds and SPL/IPL showed stronger connectivity with the bilateral dorsal seeds. Convergence with the original seeds was ensured using these as a mask to the results shown in a

Fig. 7

a Characterization of the functional differences between the networks formed by ventral (green) and dorsal (red) posterior frontal regions and their respective parietal connections. Behavioural domains are shown on top, paradigm classes on the bottom. b Characterization of the functional differences between the ventral (green) and dorsal (red) posterior (seed) frontal regions