Introduction
The objects we are facing are endowed with a special behavioral status as they offer maximal affordance for manipulation but also represent potential obstacles during locomotion. Paying a special attention to these straight-ahead elements is thus desirable. Most of the time, our gaze is directed straight-ahead, so that the important neuronal resources allocated to central vision actually process these elements efficiently. However, it can happen that an unexpected event in the surrounding space attracts attention and consequently gaze direction and central vision toward an eccentric location. Elements located straight-ahead then fall in the periphery of the retina for which vision is much less accurate. Recently, a compensatory mechanism that permits an enhanced processing of the straight-ahead direction in this case has been evidenced by single-cell recordings in the primary visual (V1) area of rhesus macaque monkeys (Durand et al.
2010). Most V1 neurons with peripheral receptive fields (RF) exhibit an increased visual sensitivity as their RF is brought closer to the straight-ahead direction by progressively shifting gaze direction. In humans, a growing body of evidences also suggests that the straight-ahead direction benefits from a privileged processing in peripheral vision. Behavioral studies (Camors et al.
2016; Durand et al.
2012) showed that participants react faster to visual stimuli if they are located straight-ahead rather than in an eccentric position, even if their visual properties (i.e. their retinal images) are strictly identical. A recent functional imaging study established that in early visual cortex (i.e. in areas V1 and V2), the blood-oxygenation-level-dependent (BOLD) response is stronger for stimuli located along the straight-ahead direction (Strappini et al.
2015). It is tempting to interpret these human results as reflecting fast and low-level neural mechanisms similar to that evidenced in macaque. However, such an interpretation is hampered by the current lack of knowledge regarding the dynamics of the straight-ahead preference in humans.
Here, we address this issue with electroencephalographic (EEG) recordings in human participants, which permits to capture the dynamics of straight-ahead facilitation with a millisecond resolution. We recorded the event-related potentials (ERPs) in response to visually identical peripheral stimuli presented either straight-ahead or at an eccentric position of the egocentric space (i.e. relatively to the body). Visual ERPs include a number of task-specific components, reflecting different cortical processing stages (Di Russo et al.
2019). In a first experiment, we adapted the design of our previous behavioral experiments with peripheral stimuli located along the horizontal meridian (Camors et al.
2016; Durand et al.
2012). However, we used here large and high-contrast stimuli that are more efficient at evoking robust ERPs. The spatial configuration of the stimuli permitted to isolate components whose latencies ranged between 70 and 350 ms after stimulus onset. Because this experiment discovered earliest straight-ahead effects at 70 ms, we performed a second experiment specifically designed to test if these early effects originated from primary visual cortex. In this case, stimuli were presented within the four visual quadrants to elicit a C1 component. This EEG component arises around 60 ms after stimulus onset and is believed to reflect the first measurable feedforward responses from primary visual areas, i.e. V1 (Clark et al.
1995; Di Russo et al.
2003) but also V2 and V3, see (Ales et al.
2010). In addition, we tested the hypothesis that lateral gaze fixation might impact pre-stimulus onset activity in posterior sites to facilitate the processing of upcoming straight-ahead visual inputs.
Discussion
The aim of this study was to characterize the dynamics of the privileged processing of the straight-ahead direction in human visual cortex. We recorded the EEG responses to peripheral stimuli that were visually identical but that were presented either straight-ahead or at eccentric positions (Fig.
1). Our experiments both demonstrated that straight-ahead stimulations lead to stronger P1, N1 and P2–P3 components (Figs.
2,
4). We did not observe any significant modifications in the latencies of these components. It suggests that the straight-ahead preference is characterized by a gain augmentation of the responses, in agreement with fMRI recordings in human (Strappini et al.
2015), single-cell recordings in macaque V1 (Durand et al.
2010) and more generally with several studies on gaze position effects on neural activations in primate, see, e.g. (Trotter and Celebrini
1999; Merriam et al.
2013). In addition, our frequency analysis of pre-stimulus brain activity suggests that the straight-ahead direction is associated with reduced alpha power in the contralateral hemisphere before stimulus onset (see Fig.
7). Previous EEG studies showed that pre-stimulus alpha power suppression is associated with enhanced evoked activity (Bacigalupo and Luck
2019; van Dijk et al.
2008) and modifications of the P1 amplitude (Fellinger et al.
2011). Our observations are also in line with the pre-stimulus preparatory activity found in the occipital areas for simple reaction tasks by Di Russo et al. (
2019). The modulation of the evoked response that we observed for straight-ahead stimuli could, therefore, be linked to this pre-stimulus and gaze related alpha suppression.
In our study, the latencies of the P1 and N1 components are in very good agreement with those measured with peripheral stimulation (Novitskiy et al.
2011; Di Russo et al.
2002). These latencies are slightly longer than those triggered by foveal stimulation, e.g. (Di Russo et al.
2019). Peripheral stimuli are indeed known to evoke slower EEG responses (Hansen et al.
2016). Because the mean wave might reflect a mixture between the C1 (peak latency of 70 ms) the contralateral P1 (peak latency of 80 ms) and the bilateral P1 (peak latency of 140 ms), it could have introduced some variability in the peak latencies we observed (see Fig.
2). To cancel out this variability, we performed additional analyses based on differences between waves. The contralateral P1 was estimated from the difference between responses to left versus right stimulations and the C1 and C2 components were estimated from the difference between upper versus lower stimulations. On the obtained difference waves, individual latencies and topographies were more consistent, which permitted a more robust characterization of the ERP components at the individual level.
Our analysis did not allow us to well discriminate between the P2 and P3 components from the mean wave, probably because they have similar topography and very close latencies in simple reaction time tasks (Di Russo et al.
2019). We, therefore, cannot be sure whether the strongest EEG responses that we obtained for these components during straight-ahead stimulation correspond to the P2 and/or the P3.
Overall, the earliest effects that we measured arose around 70 ms after stimulus onset and correspond to the contralateral occipital P1 (Fig.
3). Previous studies based on EEG source localization approaches combined with neuroimaging data proposed that this component is generated in extra-striate visual cortex, first of all in dorso-occipital areas (V3, V3A) and then in ventro-occipital regions (V4) (Di Russo et al.
2001,
2003; Martinez et al.
1999).
Our second experiment was specifically designed to determine whether this privileged processing of the straight-ahead direction can be detected in the first measurable EEG responses from early visual cortex (i.e. from areas V1, V2 and V3). We confirmed here the results of the first experiment, and, in addition, after the subtraction between ERPs to upper versus lower visual hemifield stimulations, we obtained strong C1 and C2 components (Fig.
5) whose amplitudes and latencies are very consistent with previous studies (Di Russo et al.
2003; Miller et al.
2015). With this approach, gain modulations were first observed around 130 ms after stimulus onset, which corresponds to the C2 component. We confirmed that this result was not caused by our methodological approach with an additional analysis that led to the same conclusion (Fig.
6). Because of its latency and its polarity reversal for top versus bottom stimulation, this component is believed to reflect feedbacks from higher-level visual areas to the early visual cortex (Miller et al.
2015). It was previously suggested that these feedbacks might provide information about low and high-level scene features, such as the category or the depth of the stimuli (Morgan et al.
2016; Revina et al.
2018). That top–bottom feedback signals are thought to modulate sensory input, providing information about the global scene structure. Recent models of dynamic predictive updating also proposed that these top-down signals could inform about the global structure of the visual scene (Edwards et al.
2017). Our results are in line with these studies and demonstrate that neural processing in human primary visual cortex is not purely retinotopic and that it integrates egocentric spatial properties.
In our data, the first detectable modulation of the evoked responses for straight-ahead stimuli occurred for a component whose possible cortical generators are extra-striate areas. One previous paper characterized the influence of eye position on EEG recordings using two electrodes (Andersson et al.
2004). In this study, gaze direction modulated the amplitude of the C1 component but at relatively late latencies that ranged between 94.8 and 98 ms. However, these authors only measured the responses to stimulations in the lower visual field and were, therefore, not able to compute the difference between ERPs in response to upper versus lower stimulation as in the present study. With their approach, the C1 and P1 components overlap and become difficult to disentangle (Luck
2005). Given the latency that they reported, the gaze effect in their study was most likely driven by the P1 component and is, therefore, in agreement with our results.
Our data demonstrate that straight-ahead effects become significant in early visual cortex around 130–160 ms after stimulus onset. This is compatible with previous single-cell recordings performed by our group which showed that the spike rate of V1 neurons is further increased for straight-ahead stimulations 100–150 ms after stimulus onset (Durand et al.
2010). If small incongruences may exist between the two studies (e.g. the first straight-ahead effects are measured earlier in macaque), it remains difficult to determine whether they arise from methodological and/or specie differences. Additional experiments are required to clarify this point, for example using intracranial recordings in implanted epileptic patients, see (de Jong et al.
2016).
Can the straight-ahead preference be related to attention mechanisms? In our previous behavioral study (Durand et al.
2012), a dual task at fixation did not alter the straight-ahead effects. Gain modulation for straight-ahead stimulations was also observed in an fMRI study performed under passive viewing conditions (Strappini et al.
2015). However, these manipulations did not require full attentional resources and, therefore, do not rule out an attentional explanation (see the discussion in Strappini et al. (
2015). Indeed, several ERP experiments showed that the P1, N1 and P2 components are affected by spatial attention (Hillyard and Anllo-Vento
1998). Van Voorhis and Hillyard notably suggested that these attention effects could arise in the early phase of the P1 at a latency (65 ms after stimulus onset) that is consistent with the timing of our earliest straight-ahead effect on the evoked responses (at 68 ms) (Van Voorhis and Hillyard
1977). Moreover, a growing number of studies proposed that spatial attention leads to alpha desynchronization (i.e. to a decrease in alpha power) in the ‘attended’ hemisphere (see e.g. Kelly et al. (
2006), as the straight-ahead direction in our study. Finally, before stimulus onset, we also observed consistent gaze-dependent drifts in frontal cortex that might correspond to pro-active attention mechanisms (see Supplementary Fig. 3 and the accompanying text). Altogether, these findings suggest that the privileged processing of the straight-ahead direction might be driven by neural mechanisms close to those involved in spatial attention and possibly triggered by proprioceptive signals related to the position of the eyes in their orbit.
In a recent study, we showed that straight-ahead stimuli triggered faster saccades than stimuli in eccentric space (Camors et al.
2016). The straight-ahead amplification observed for the P2 component in the present study is in line with a cortical mechanism that would facilitate action preparation. Indeed, besides other functions, an important process that affects the P2 amplitude is cross modal visuo-tactile integration. Vision of the hand enhances the tactile P2 (Torta et al.
2015) and the vertex P2 potential is also stronger for congruent visuo-tactile events (Longo et al.
2012). Straight-ahead effects on the P2 could, therefore, in addition to attention mechanisms, could reflect visuo-motor integration that facilitates interaction with objects located in front of the body.
By contrast with our previous report (Durand et al.
2012), we did not find significantly shorter reaction times for straight-ahead stimuli in the present study. Overall, our reaction times were significantly longer (350 ms against 300 ms), despite the fact that we used larger (~ 6°*10° against 2° in diameter) and more contrasted (100% against 30%) stimuli to evoke strong ERP components. Both the extension of the reaction times and the absence of a straight-ahead effect might have their root in the fact that the stimuli were located within the peri-personal space in our previous study (40 cm), while they were farther away (150 cm) in the present experiments. Since peri-personal space is known to be associated with behavioral facilitation (Graziano and Cooke
2006) and to trigger shorter reaction times (Li et al.
2011), we hypothesize that the behavioral straight-ahead facilitation measured by simple reaction time might occur only within the peri-personal space.
When presented in eccentric position, stimuli had small vertical disparity because they were closer to one eye than to the other. In our previous studies (Durand et al.
2010,
2012), we showed that under monocular viewing, straight-ahead effects remained unchanged. Moreover, vertical disparities become negligible at long viewing distances such as the one used in this study. It is, therefore, unlikely that vertical disparity had an impact on our results. Poor gaze fixation with a drift towards the center could be another confound, but we controlled gaze fixation with an eye-tracker in a subset of subjects and did not find any significant differences between fixation during our straight-ahead and eccentric conditions (see supplementary materials file). Such a centripetal bias would also modify the retinotopic positions of our stimuli and alter the C1 but we did not measure such a change. Finally, we used a curved screen so that straight-ahead and eccentric stimuli were localized at the same distance from the subjects. It excludes distance as the cause of our results.
To conclude, our EEG data confirmed previous reports showing that the straight-ahead direction is preferentially processed in the humans (Strappini et al.
2015) as it is the case in non-human primates (Durand et al.
2010). The earliest measurable effects on the evoked responses appeared around 70 ms after stimulus onset. Alpha power suppression in the hemisphere contralateral to the straight-ahead direction was also observed before stimulus onset. Altogether, the neural mechanisms described here demonstrate that visual processing, even in its early phases, is not uniquely retinotopic and also reflects egocentric properties. They, therefore, impose important modifications on current models of vision and spatial perception.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.