In the central visual system feedback interactions parallel the feed forward connections and there are arguments for considering its function in terms of iterative interactions in a circuit rather than a sequence of processing steps. From this viewpoint it is notable that in the primate visual system the cortical motion area MT/V5 provides feedback to the primary visual cortex (V1) with the potential to provide cascaded feedback to thalamic relay cells in the LGN (Born & Bradley 2005, Sillito et al. 2006, Briggs & Usrey 2009). Feedback from MT to V1 exerts strong and clear effects on V1 responses to flashing and moving stimuli. A component of feedback connections from MT to V1 terminate in upper and lower layer 6 where cortico-geniculate neurons providing feedback to the magno and parvo cellular layers of the LGN are found. Layer 6 cells in V1 influence the visual responses and response pattern of LGN cells. This has the surprising implication that information about stimulus motion and direction reflected in the responses of MT cells is available to LGN cells. Is it? To date there is no evidence to suggest that LGN cell visual response properties resemble those of MT cells. However this might reflect the fact that cascaded feedback from MT direction columns is pooled so that for any given point in visual space a complete set of direction columns influences the responses of LGN cells, in effect averaging out any directional bias. To isolate possible influences from MT we explored the effect of blocking the activity of a locus in the columnar organization of MT representing a particular direction of motion with a brief iontophoretic pulse of GABA. We predicted that if feedback from MT influenced LGN cell responses blocking the MT locus would cause changes in the responses of LGN cells biased to that direction of motion. We recorded from 133 primate LGN cells (Macaca mulatta, anesthetized with sufentanil 4μg/kg/hr supplemented with halothane (0.1-0.4%) in 70% N2O/30% O2. Neuromuscular blockade was induced with 0.1mg/kg/hr vecuronium bromide and anaesthesia monitored as previously described (Jones et al. 2001) during reversible blockade of an MT direction column. Of these LGN cells, 96 showed a reversible and significant shift (P <0.05 paired two tailed T test) in directional bias during the focal blockade of MT. There was a significant difference between LGN cell control firing rate to stimuli moving in the preferred direction of the MT column and those observed when the activity of the MT column was blocked by drug application (P = 0.023450 Wilcoxon matched pairs test) but not between firing rates for control and drug conditions when the stimulus was moving in the non preferred direction. We checked whether there was variation in the effect of MT inactivation across the three types of LGN cells. The geometric mean of the change in directional bias was 43.9% for magnocellular cells (n = 24, 95% CI 37.07, 51.89), 19.8% for parvocellular cells (n = 92, 95% CI 15.15, 25.90) and 6.3% for konicellular cells (n = 17, 95% CI 2.75, 14.50). These values across the three cell classes were significantly different to each other (P = 0.0000, Kruskal-Wallis Anova). We checked if the magnitude of effect was influenced by whether the LGN fields overlapped with the MT field or not. For parvocellular cells there was a significant variation (P = 0.000002 Mann Whitney U test) in effect if the LGN receptive field overlapped the MT field or not, with geometric mean values for overlapping cells of 9.9% (n = 34, 95% CI 6.19, 15.71) and non overlapping cells of 29.8% (n = 58, 95% CI 22.40, 39.69). The difference in values for overlapping and non-overlapping koni and magnocellular cells were not significant. We also observed larger effects in magno and parvocellular cells if their receptive fields lay close to the angular vector defining the preferred direction of motion of the MT column inactivated. Basically our observations reveal a constant stimulus driven background modulation of LGN cell responses by feedback from MT. This introduces the context of salient motion integrated over a much wider area than the subcortical process in isolation. In the sense that these changes in LGN cells responses will provoke a change in the input to V1, and thus a change in the input to MT and a consequential change in MT responses and its feedback back to V1 and the LGN, our data argue for a constant reiterative mechanism locking onto the stimulus. From this perspective the representation of the stimulus sits in the interaction between the levels.