Isoluminant coloured stimuli are undetectable in blindsight even when they move
- First Online:
- Cite this article as:
- Alexander, I. & Cowey, A. Exp Brain Res (2013) 225: 147. doi:10.1007/s00221-012-3355-6
- 298 Views
Moving stimuli are the most effective of all in eliciting blindsight. The detection of static luminance-matched coloured stimuli is negligible or even impossible in blindsight. However, moving coloured stimuli on an achromatic background have not been tested. We therefore tested two blindsighted hemianopes, one of them highly experienced and the other much less so, to determine whether they could perform what should be one of the simplest of all motion tasks: detecting when an array of coloured stimuli moves. On each trial, they were presented in the hemianopic field with an array of spots, all red or green or blue or achromatic, in a circular window and on a white surround. The spots moved coherently in the first or second of two short intervals. The subject had to indicate the interval in which the motion had occurred. The luminance of the spots was varied across different blocks of trials, but the background luminance remained the same throughout. For each colour, there was a ratio of luminance between the spots and the white surround at which performance was not significantly better than chance, although at other ratios, performance was good to excellent, with the exception of blue spots in one subject. We conclude that detecting global coherent motion in blindsight is impossible when it is based on chromatic contrast alone.
Blindsight is the ability of patients with destruction or denervation of the striate cortex (V1) to detect, discriminate and even identify visual stimuli confined to their clinically blind-field defects and which they deny seeing (for review see Cowey 2010). What they mean by “not seeing” is often debated but this interesting question is not considered in the present paper. Instead, we concentrate on the evidence, first provided by Riddoch (1917) at a time when there was no brain imagery or neurohistology that could confirm the total destruction regions of V1, that residual visual sensitivity in the field defect is most easily demonstrated with fast-moving stimuli (Azzopardi and Cowey 2001; Barbur et al. 1993; Zeki and Ffytche 1998). The explanation often given for this apparent sparing of motion detection is that the extra-striate cortical motion complex (V5/MT) continues to receive visual input in the absence or reversible inactivation of V1 (Azzopardi et al. 2003; Barbur et al. 1993; Ffytche et al. 1995; Rodman et al. 1989; Girard et al. 1992) and that one of the afferent routes is via surviving K-cell neurons in the interlaminar layers of the otherwise degenerated dorsal lateral geniculate nucleus of the thalamus (Schmid et al. 2010). Since the interlaminar koniocellular layers receive retinal input from, amongst others, S-cone positive retinal ganglion cells (Brett et al. 2008; Szmajda et al. 2008; Roy et al. 2009) and there is a direct projection from these layers to area V5/MT (Sincich et al. 2004), they should contribute to the activity in V5/MT that survives destruction of V1 and might even convey information about wavelength. However, Gegenfurtner et al. (1994) reported that neurons in area MT of normal macaques that respond to isoluminant motion do so in a such manner that they are unlikely to be the source of chromatic motion processing. We examined the possibility of chromatic processing in blindsight by presenting random dot kinematograms that varied in colour and luminance contrast in the blind field of two human hemianopes whose other blindsight properties have already been thoroughly investigated.
The two hemianopes were GY and MS, whose cortical lesions and blindsight have been described elsewhere (e.g. Barbur et al. 1980, 1993; Brent et al. 1994; Azzopardi and Cowey 2001; Bridge et al. 2008; Alexander and Cowey 2009 for GY, and Heywood and Cowey 2003; Pavan et al. 2011; Cowey et al. 2011 for MS). In brief, GY suffered a unilateral medial occipital cortex lesion caused by a road traffic accident when he was 8 years old. The occipital pole is the only part of his left striate cortex to survive, accounting for his macular sparing of about 3o. MS contracted herpes encephalitis in 1971 which destroyed most of his ventral temporal cortex of both hemispheres as well as the calcarine cortex on the right, leaving him with complete left homonymous hemianopia. The surviving hemifield is agnostic for faces and objects and has total achromatopsia. At the time of the current investigation, GY was aged 59 and MS 61. Both subjects gave their informed consent prior to their inclusion in the study, which was performed in accordance with the ethical standards of the 1964 Declaration of Helsinki.
Two male control subjects, SR aged 61 and AS aged 60 with normal or corrected vision also took part.
Stimuli were generated by a Dell computer with a 3-GHz Pentium processor using Matlab Psychtoolbox 2007. They were displayed on a Viglen Ency 19TF monitor with a frame rate of 75 Hz. Moving stimuli were presented at 32o/s along the horizontal axis. Each block contained 60 trials and subjects completed one or occasionally two blocks for each condition.
The stimulus was a random dot kinematogram with 100 % coherence, viewed through a circular 20° window cut out of black card and placed over the VDU. Each window contained on average 30, 1-degree spots in a uniform 10 cd/m2 white surround, with the exception of blue spots where the surround was 5 cd/m2. The spots were always present on the screen but moved in one of two intervals. Only in the inter-trial interval were the positions of the spots re-drawn for the next trial. The spots had an infinite lifetime, meaning that every spot moved smoothly throughout each trial. The fixation point was a conspicuous white spot on the black card 10° to one or other side of the lateral edge of the circular window, meaning that the nearest edge of the window was 10° into the hemianopic field. There were 22 different stimulus spot conditions; blue (3, 5, 8, 10, 20 cd/m2), green (6, 8, 10, 13, 75 cd/m2), red (6, 8, 10, 13, 75 cd/m2) and achromatic (0, 6, 8, 11.5, 13, 75 and 180 cd/m2) and each block of 60 trials contained only one stimulus type. Luminance was measured with a Minolta Chroma Meter, CS-100. The white surround was 10 cd/m2 for the red, green or achromatic spots but was only 5 cd/m2 for the blue spots. The latter was because the maximum attainable luminance of blue spots was lower than that of the other colours and maximum stimulus contrast therefore required a dimmer white surround.
Subjects sat in a dimly lit room with the head supported on a chin rest 57 cm from the display monitor. Fixation was continuously monitored using an SMI RED-11 infra-red eye tracker (Sensorimotoric Instruments). Trials where fixation strayed by more than about 2–3° during stimulus presentation were eliminated but this rarely occurred (less than 2 %). Viewing was binocular and both subjects wore their near-vision spectacles. Trials were initiated by pressing the space bar. The subject responded verbally to indicate whether the spots moved in the first or second of two brief intervals and the experimenter entered the response on the keyboard. Subjects initially carried out a few practice trials in their ‘seeing’ field in order to determine that they understood the task. The coherent global motion was always to the right (away from the fixation point for GY who is a right hemianope and SR and towards the fixation for MS who is a left hemianope, and AS). Each trial consisted of two 300 ms intervals, separated by 1 s, and the start of each interval was marked by a 40-ms beep.
The control subjects scored 100 % correct on all the conditions.
Blue: Binomial analysis demonstrated that GY’s performance was significantly above chance when the drifting spots were 3 or 10 cd/m2 (p < 0.001) but not when they were five (p = 0.927) or 8 (p = 0.523) cd/m2 (Fig. 1, top left).
Green: His performance was significantly above chance when the stimuli were 6, 8, or 75 cd/m2 (p < 0.001), but not when they were 13 cd/m2 (p = 0.120). At 10 cd/m2, he was paradoxically significantly worse than chance (p = 0.008), Fig 1, bottom left.
Red: GY scored significantly above chance (p < 0.001 in all cases), except at 8 and 10 cd/m2 isoluminance (p > 0.05) (Fig. 1, top right).
Achromatic: Fig. 1, bottom right shows that performance was not significantly different from chance when the stimuli were the same luminance as the surround (p = 0.120) and therefore invisible, whereas he scored well above chance levels (p = 0.000 in all cases) for all dimmer or brighter stimuli.
Overall, no matter what the colour of the coherently moving spots against a white background, there was a luminance ratio, close to photometric isoluminance, where the detection of the coherent motion became impossible. Nor was there any clear evidence that any colour was better than the others, with the exception of 13 cd/m2 stimuli where his score was not better than expected by chance with the green stimuli but better than 90 % correct for the red and achromatic stimuli.
Moving stimuli was much more impaired than GY. With blue stimuli, he was close to chance whatever their luminance (p > 0.05 in all cases). He scored better than 70 % correct with the dimmest green and just under 70 % with the dimmest red and better than 70 % correct with the brightest green and red (p < 0.001 in all cases). The latter were almost identical to his scores with the brightest white (p < 0.001). MS scored highly on the dimmest (0 cd/m2) and the two brightest (75 and 180 cd/m2) achromatic stimuli (p < 0.001 in all cases), whereas his other scores were not significantly different from chance (p > 0.05 in all cases). For MS, his performance reflected the luminance contrast of the moving stimuli, whether green, red or white. Blue spots were entirely ineffective.
Throughout testing, no information was provided about how well the subject was performing until after the end of each block of trials. And before revealing the score to GY and MS, they were always asked what—if anything—they experienced during that block. MS found it difficult to deal with this request. He said that he was just doing his best by guessing (blindsight type 1), but with the dimmest and brightest stimuli, where he performed better than expected by chance, he “thought that there might have been something there”. But he could not describe it as anything more than “a feeling” (blindsight type 2). He never admitted to seeing anything. GY found it easier to describe his experience. Despite never experiencing anything like a visual percept, he was “confident” at the higher or lower luminance contrasts that he was scoring well because he “knew” in which interval the display moved. When his score was better than 90 % correct, he reported that he was “aware of something on nearly every trial”. But on further questioning, he described that awareness as “knowing when the display moved or being confident about his judgement” or both, even though he did not see anything.
Cone contrasts for the stimuli used. The stimuli are listed in the first column
In the normal, seeing, hemifield both subjects could perceive and describe the motion when it was shown to them for a few trials. This was even true for MS, who is achromatopsic in his seeing hemifield. However, although he can detect the movement, he cannot discriminate its colour. The blind-field results show that in the absence of V1, there is no selective sparing of even the simplest possible colour motion via the k-cell projection from the interlaminar layers of the otherwise degenerated dLGN to the cortical visual motion complex MT/V5 even though in the undamaged brain motion from hue activates area MT/V5 (Ffytche et al. 1995). However, as mentioned above, Gegenfurtner et al. (1994) reported that those cells in MT that may respond at isoluminance are unlikely candidates for chromatic processing. We predicted that the hemianopes would be able to detect the coherent motion of the blue spots at any ratio of luminance via the spared S-cone blue/yellow system but they could not. This does not imply that the k-cell pathway, which abundant other evidence indicates is particularly involved in blue-yellow S−(L+M) colour opponency, does not normally transmit information about colour motion via the interlaminar layers of the LGN. It is more likely that in the absence of V1 which normally provides, directly or indirectly, the major input to the cortical motion complex V5/MT+, the latter has severely impaired properties despite retaining some of its normal afferent input (e.g. Azzopardi et al. 2003).
Although performance with green spots was poorer at the smallest luminance contrasts than with blue, red or white in GY the differences were not striking. And the likely explanation is that the green spots are less effective at providing contrast in blind fields because the spectral sensitivity of surviving retinal M cells is too close to that of the white surround. Nor was there any strong indication of better (or worse) performance at similar luminance ratios with white as opposed to coloured spots. In other words, the chief factor in determining motion detection was luminance contrast. However, performance of GY was so good at the higher contrast ratios (Fig. 1) that any differences might have been obscured by ceiling effects.
Figure 1 shows that GY performed much better than MS. In fact, MS was unable to perform above chance levels with blue spots at any luminance ratio. The explanation almost certainly arises from the difference in their cortical damage. GY has a lesion that involves only V1 and V2 and perhaps a little of V3. Some of his other cortical visual areas are—at least grossly—intact and excitable by a variety of visual stimuli (Barbur et al. 1993; Holliday et al. 1997; Baseler et al. 1999; Morland et al. 2004). But the cortex contralateral to MS’s hemianopia is extensively destroyed along the entire ventral visual processing stream from V1 to rostral inferior temporal cortex (Heywood et al. 1991; Cowey et al. 2008). His impoverished motion detection in the blind hemifield presumably depends entirely on the midbrain projections to the dorsal, parietal, pathway. Indeed, Schenz and Zihl (1997a, b) reported impaired visual motion processing after damage to parietal brain structures involved in motion and form.
Why did GY score only 45/120 when the green spots were isoluminant with the surround? He claimed that he was unaware of anything on those trials and that as the task was impossible he responded randomly. Given the number of binomial tests performed at or close to isoluminance (10), the paradoxical result of performing at worse than chance with isoluminant green spots becomes insignificant after correcting for multiple comparisons.
Guo et al. (1998), using forced-choice and rating methods, observed that GY scored above chance on discriminations with chromatic stimuli on an isoluminant chromatic background and concluded GY is able to use wavelength and luminance information in his blind field which appears to conflict with our conclusions. However, we think not. Guo et al. did not control for screen artefacts that are common at the left edge of raster displays or for artefacts that can arise in the left side of viewing goggles. When Azzopardi and Cowey (see Cowey 2010) used similar viewing conditions with GY, he detected both types of artefact but could not discriminate the targets in his blind field once they were removed. This also explains why in the experiment by Guo et al. GY reported seeing the stimuli and their colour. In the present experiment, GY denied ever seeing any of the displays. Whether other factors might produce different results from those reported here is unknown and would require extensive further investigation. For example, would red spots on a blue background be much more detectable; would different directions of coherent motion improve performance; might detection improve at lower or higher speeds? Unfortunately, there are no strong theoretical reasons to expect that such changes would affect the result obtained here that luminance rather than wavelength is the most important factor.
Our research was supported by an EPA Cephalosporin Trust Grant to IA and a Leverhulme Emeritus Fellowship to AC. We thank the hemianopic subjects and the sighted control subjects for their willingness to carry out many hours of psychophysical testing.