Exp Brain Res (2003) 151:536–541 DOI 10.1007/s00221-003-1509-2
RESEARCH ARTICLE
J. Peter Maurer · Michael Bach
Isolating motion responses in visual evoked potentials by preadapting flicker-sensitive mechanisms Received: 14 August 2002 / Accepted: 22 April 2003 / Published online: 8 July 2003 Springer-Verlag 2003
Abstract Onset of visual motion evokes a component in the EEG, the motion onset VEP. Exploring its motion specificity with a direction-specific adaptation paradigm, previous work demonstrated that less than 50% of the motion onset VEP represents actual motion detection. Here, we tested whether preadaptation of flicker-sensitive mechanisms can help to isolate motion-specific responses in the VEP. Flicker preadaptation was accomplished by limiting dot lifetime in the random-dot kinematograms that we used to study the direction specificity of motion adaptation. With unlimited dot lifetime, motion adaptation reduced the VEP amplitude to 35% (adapted direction) and 50% (opposite direction). With the shortest dot lifetime (40 ms), motion adaptation reduced the amplitude to 55% (adapted direction) and 70% (opposite direction). These findings suggest that random-dot kinematograms with short dot lifetimes could improve the investigation of human motion processing, be it in electrophysiology or other fields. While such stimuli successfully preadapt flicker-related components, they still evoke a sizable response, of which an estimated 70% is motion-specific. Keywords Motion detection · Adaptation · Cortex · Human · VEP
Introduction Visual motion detection has been investigated extensively in single-cell studies (macaque monkey: Albright 1984; Britten and Newsome 1998; wallaby: Ibbotson et al. 1998, 1999; Treue et al. 2000) and has thereby been localized in certain cortical areas such as V1, V2, V3, and V5 (MT) in J. P. Maurer · M. Bach ()) Elektrophysiologisches Labor, Universitts-Augenklinik, Killianstr. 5, 79106 Freiburg, Germany e-mail:
[email protected] Tel.: +49-761-2704060 Fax: +49-761-2704052
macaque monkeys (as reviewed by DeYoe and Van Essen 1988), the decisive criterion being direction specificity (Borst and Egelhaaf 1989). In humans, direction specificity cannot be assessed directly, as the instruments at our disposal, such as psychophysics (e.g., Levinson and Sekuler 1975, 1980; Raymond 1993), electrophysiology (namely visual evoked potentials: e.g., MacKay and Rietveld 1968; Clarke 1972, 1973a, 1973b; Kubov et al. 1990; Kuba and Kubov 1992; Snowden et al. 1995), and imaging methods (PET: Zeki et al. 1991; fMRI: Tootell et al. 1995; He et al. 1998; Culham et al. 1999; Sunaert et al. 1999; Hautzel et al. 2001), do not have the spatial resolution to scrutinize motion-specific mechanisms on a single-cell level. Data thus obtained represent the response of cells with all kinds of preferred stimuli pooled in one signal, possibly even including cells that are not employed in motion processing at all. This dilemma can be solved by taking advantage of the fact that motion detectors not only show a directionspecific response but also direction-specific adaptation behavior (Levinson and Sekuler 1980; Raymond 1993). Using a motion adaptation technique, Bach and Hoffmann (2000) and Hoffmann et al. (2001) found that the dominating negativity in the motion onset VEP at occipital and occipitotemporal sites around 140–200 ms after motion onset (N2: Gpfert et al. 1983; Mller et al. 1985; Bach and Ullrich 1994) indeed reflects the activity of direction-specific units as, after motion adaptation, its amplitude is significantly more reduced when testing in the adapting direction than when testing in other directions, particularly the opposite direction. However, the N2 amplitude reduction is only partially due to motionspecific mechanisms. There is an additional directionindependent amplitude reduction accounting for at least half the effect, which can also be induced by pattern reversal and even pattern onset adaptation. Hence, Hoffmann et al. (2001) concluded that this ’global effect’ (Bach et al. 1996) is associated with the adaptation of the ubiquitous temporal channels, i.e., mechanisms susceptible to local luminance modulations (flicker) inherent in any motion stimulus.
537
The aim of the present study is to enhance the specificity of evoked potentials for reflecting motion components in the stimulus. One way to minimize the contribution of flicker onset would be to analyze responses to motion reversal (e.g., Clarke 1972, 1973b, 1974; Odom et al. 1998). These responses, however, do not allow distinguishing between motion onset and offset. Therefore, we used motion onset responses in conjunction with a double adaptation paradigm: the first adaptation stage (constantly flickering stimuli) targets flicker-sensitive mechanisms, while the second stage targets flickersensitive mechanisms and, in addition, adapts a specific motion direction. Indeed we found that motion responses in this doubly adapted condition were more directionspecific than without the first adaptation step, improving the value of the motion VEP as a tool to further our understanding of human motion perception and its possible malfunction.
Materials and methods Subjects Eight human observers with normal or corrected-to-normal visual acuity (1.0 decimal acuity) participated in the study; five of the observers were naive as to the experimental question. All subjects gave their informed written consent to participate in the experiments. All experiments were conducted in accordance with the Declaration of Helsinki (World Medical Association 2000).
Fig. 1 Stimulus pattern. Dots moved coherently with a speed 8.6/s in front of a darker background; the Michelson contrast was 10%. A single dot’s lifetime was limited to 40, 133, or 413 ms, or was unlimited. Subjects were instructed to fixate the centrally placed target
Stimuli Stimuli were generated by a Power Macintosh G4 computer (Bach 1999) and presented on a CRT with a frame rate of 75 Hz at a viewing distance of 57 cm. The stimulus pattern (Fig. 1) consisted of 800 randomly positioned dots (size 0.25 0.25) moving coherently in front of a darker background within a stimulus aperture of 30 diameter. The speed was 8.6/s, and the Michelson contrast of the dots versus the background was 10%. This contrast was set to preferentially stimulate the magnocellular system (Kaplan and Shapley 1986; Bach and Ullrich 1997). Over the outer 2 of the stimulus, the luminance of the dots was reduced, resulting in a Michelson contrast of 4.8%. Space averaged mean luminance of the pattern was 74 cd/m2. To reduce optokinetic nystagmus, a bright cross in front of a dark disc (2.5 diameter) was centered on the pattern. In previous experiments with similar stimuli this kind of fixation target has been shown to sufficiently extenuate tracking eye movements (Bach and Hoffmann 2000; Hoffmann et al. 2001; Hoffmann and Bach 2002). One stimulus trial (overall duration 2,400 ms) consisted of three epochs (Fig. 2): 2,000 ms adaptation, 200 ms stationary pattern, and 200 ms motion (‘test’; to the right or to the left). During the 2,000ms adaptation epoch, the pattern remained either stationary (‘baseline condition’), or moved to the right (‘adaptation condition’). Stimulus trials were presented consecutively in a cyclic design, providing for a continuous, deep adaptation state (Hoffmann et al. 1999). Throughout all three stimulus epochs a single dot’s lifetime ‘td’ was either limited (cf. Baker et al. 1991; Burr and Santoro 2001) to a certain mean dot lifetime (td=40 ms, td=133 ms, or td=413 ms; actual lifetimes were equally distributed between 0 ms and the specified td2) or was not limited at all (td=1). Dots that had either consumed their lifetime or moved out of the aperture reincarnated at a random position within the aperture. Subjects perceived a stationary or moving pattern with various amounts of superimposed flicker, except for the unlimited dot lifetime stimuli.
Fig. 2 Stimulation scheme. One stimulus trial (overall duration 2,400 ms) consisted of three epochs: 2,000 ms adaptation, 200 ms stationary pattern, and 200 ms motion (‘test’; motion to the right or to the left). During the 2,000-ms adaptation epoch, the pattern remained either stationary (‘baseline’ condition) or moved to the right (‘adaptation’ condition), cf. Table 1 Scase et al. (1996) have found similar psychophysical coherence thresholds for various kinds of noise degradation of motion stimuli (random position, random walk, and random direction). Thus, we surmise that our limited-lifetime stimuli (corresponding to Scase et al.’s random position stimuli) are equally suitable motion stimuli as other noise-degraded motion stimuli. VEP and EOG recordings VEPs were recorded from three derivations: Oz (American Encephalographic Society 1994), Otr, and Otl (5 cm right and left from Oz, respectively), referenced to averaged ears. The ground electrode was attached to the right wrist. Approximately 100 trials were averaged for each motion direction and dot lifetime. Signals were amplified, filtered (first-order bandpass, 0.3–70 Hz; Toennies ‘Physiologic Amplifier’), and digitized at a sampling frequency of 500 Hz with a Power Macintosh 7100 computer. Vertical eye movements and blinks were recorded with supra- and infraorbital electrodes at one eye.
538 Table 1 The experiment was subdivided into 16 counterbalanced blocks, differing in dot lifetime (td) and adaptation state (column ‘Condition’) Block
td(ms)
Condition
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
40 40 133 133 413 413 Unlimited Unlimited Unlimited Unlimited 413 413 133 133 40 40
Baseline Adapted Baseline Adapted Baseline Adapted Baseline Adapted Adapted Baseline Adapted Baseline Adapted Baseline Adapted Baseline
Results Figure 3 displays grand mean motion onset VEP traces for both stimulus directions, averaged across all subjects. The baseline condition yields typical motion onset VEPs, dominated by the N2; their amplitude decreases with shorter dot lifetime (td). Motion adaptation reduces the N2 amplitudes depending on stimulus direction. The N2 amplitude reduction is maximal for identical adaptation and test directions at both Oz (compare Fig. 3A with 3B) and Ot* (compare Fig. 3C with 3D), demonstrating the direction-specific effect of motion adaptation. When comparing the baseline/adaptation amplitude ratios between the left and the right column, the increasing
Procedure Stimuli were presented in a counterbalanced blocked design (Table 1). Each block consisted of 50 trials with test direction right and 50 trials with test direction left. Within each block, trials (i.e., test directions) were arranged in a random order. To avoid adaptation ’carry-over’ between blocks, there were 2-min breaks at every transition from adaptation to baseline, or longer if subjects still perceived a motion after-effect. Data analysis Trials were analyzed off-line over the interval from 100 ms before motion onset to 700 ms after motion onset. Trials with blinks, detected with a threshold criterion of €100 mV, were discarded. Averaged sweeps were digitally filtered (0–20 Hz). To counter interindividual variations, the records were normalized (by the rationale of multiplicatively normalizing EEG data; see Klistorner et al. 2000). The mean value from 100 ms before to 70 ms after stimulus onset of the averaged trace was used as zero reference for peak measurements. Since motion onset potentials are often strongly lateralized (Andreassi and Juszczak 1982), we attempted to optimize the signal-to-noise ratio of the N2 amplitudes by introducing the hybrid derivation Ot*, covering recordings from either Otr or Otl. This selection of recording site (Otr or Otl) depended on which of them yielded the greater N2 amplitude (i.e., the greater mean of N2 peaks in response to the unlimited dot lifetime baseline stimuli) and was fixed for each subject throughout all stimulus conditions. N2 was maximal at the right derivation in five out of eight subjects and maximal at the left derivation in the remaining three subjects. Statistics The statistical significance of experimental effects was assessed with a repeated measures ANOVA (independent factor: electrode site; within factors: motion direction, motion adaptation, dot lifetime). The effect of motion adaptation was assessed in eight post hoc paired t-tests (shortest/longest dot lifetime adapting/nonadapting direction electrode site Oz/Ot*). Thus, the Bonferronicorrected significance level was a=0.05 / 8 = 0.00625.
Fig. 3A–D Grand mean VEP traces (across eight subjects) for both stimulus directions at two recording sites (Oz: A, B; Ot*: C, D) without motion adaptation (baseline, solid trace) and after adaptation to rightwards motion (dashed trace). The direction of the test stimuli is given at the top; recording sites are indicated on the left. For each stimulus direction/recording site combination, four sets of traces are shown, representing four different dot lifetimes (td). Thin lines depict the standard error of the mean. The motion onset VEP is dominated by N2 at about 170 ms. Due to stronger flicker preadaptation, baseline N2 amplitudes are smaller with reduced dot lifetime. After motion adaptation, N2 amplitudes decrease depending on stimulus direction; the N2 amplitude reduction is maximal for identical adaptation and test direction (A, C), demonstrating the direction-specific effect of motion adaptation
539
however, the greater the ratio difference (see above) between both motion directions. At Ot*, for the minimum dot lifetime of 40 ms, the adapted N2 amplitude for the non-motion-adapted test direction (leftwards, Fig. 4D) does not significantly differ from the baseline N2 amplitude (baseline/adaptation amplitude ratio=0.7; P=0.1), while it is still at approximately 50% (i.e., baseline/adaptation amplitude ratio=0.5; significantly different, P
