The cerebral cortex has a marked propensity for generating gamma oscillations (30–100 Hz) in behaving mammals, including humans (Steriade, 2006). While there is no consensus on their function, which may encompass fundamental cognitive tasks (Fries et al., 2007), the cellular mechanisms producing these rhythms are relatively well understood. Gamma oscillations are an emergent property of highly interconnected networks of inhibitory interneurons, and their period is mainly dictated by the time course of GABAergic synaptic potentials (Bartos et al., 2007).

The basal ganglia are essential for motor control and action selection (Gurney et al., 2004). The operation of this system is intimately linked to that of the cortex, from which it receives its main input, and to which it projects back via the thalamus. The massive cortical input to the basal ganglia is mainly directed to the striatum. This nucleus, with its complex circuitry, is a major information processing centre for this system. Furthermore, dopamine-dependent corticostriatal synaptic plasticity (Calabresi et al., 2007) is a cellular substrate for reward-mediated learning.

Unlike the subthalamic nucleus/globus pallidus networks (Bevan et al., 2002), the striatum does not appear to be autonomously rhythmogenic. Nevertheless, it makes sense to expect that cortical oscillations will exert a major influence on striatal activity. However, the corticostriatal mapping is highly complex, as orderly topographical projections coexist with brisk discontinuities (Brown et al., 1998). Thus, it is difficult to predict how gamma oscillations produced by different cortical areas will affect the striatum. In this issue, Berke (2009) shows that, in behaving rats, gamma oscillations in the ventral striatum are driven both by the piriform cortex (which tends to oscillate at approximately 50 Hz), and the frontal cortex (which tends to oscillate at 80–100 Hz).

The balance of power between these rhythms shifts dramatically in favour of the faster one when the animals receive a reward, or when they are under the influence of dopaminergic stimulants. It is not clear how the frontal cortex takes control of ventral striatal circuits under these conditions; yet, it is intriguing that ventral striatal networks apparently switch from a regime in which sensory inputs are preponderant to one dominated by cognitive signals, when attentive and motivational drives exceed a certain level.

Another intriguing and unexpected finding is that it is the fast-spiking interneurons, not the projection neurons, that show significant entrainment to the cortical rhythms. Furthermore, some of these interneurons (that mediate feed-forward inhibition in the striatum) are selectively entrained by either piriform or frontal cortical oscillations. This provides further evidence that, despite being connected by gap junctions (Galarreta & Hestrin, 2001), fast-spiking interneurons fire in a highly individual manner (Berke, 2008).

These observations raise important questions. Do fast-spiking interneurons receive cortical inputs that are substantially different from those of projection neurons? Do electrically connected interneurons create multiple ‘syncytia’ driven by different cortical areas? Finally – and most importantly – what is the functional significance of striatal gamma oscillations? Berke’s findings will undoubtedly encourage new investigations aimed at tackling these issues.


  1. Top of page
  2. Refernces