All procedures were carried out in accordance with the European Communities Council Directive of 24 November 1986 (86/609/EEC) and with the approval of the Animal Welfare and Ethical Review Body at the University of Nottingham, UK. Experiments were conducted on 15 male and female pigmented GPs weighing 300–500 g at the onset of behavioral testing. GPs were bred in-house and group-housed on a 12:12 h light:dark cycle. Food and water were freely available.

The behavioral method used to identify tinnitus in this study is based on a gap detection paradigm devised by Turner et al. (6) in which we measured flexion of the pinna, or the Preyer reflex (7). The magnitude of the Preyer reflex is calculated as pinna displacement under different acoustic conditions, and these measurements are used to quantify gap-induced PPI of the reflex. In animals experiencing tinnitus, PPI is compromised; hence, the paradigm allows objective identification of tinnitus. This method is described in detail elsewhere (7, 8).

Baseline PPI of the Preyer reflex was measured in each GP over a 2-week period (minimum of three and a maximum of six testing sessions). Startling stimuli [broadband noise (BBN) bursts of 20 ms; rise/fall time of 1 ms] and continuous background noise conditions [either BBN or 2 kHz wide narrow-band noise (NBN) centered at 5, 9, 13, or 17 kHz] were used, as described previously (7, 8). The gap used to elicit gap-induced PPI in the present study was 50 ms in duration, consistent with that used by others [e.g., Ref. (6, 12, 13)]. After 2 weeks of baseline data collection, the significance of PPI was calculated as described in Berger et al. (7). Briefly, baseline data were pooled and the differences between “no gap” and “gap” trials were tested for statistical significance using a Wilcoxon rank-sum test to a 95% confidence rating for each GP at each background frequency. GPs that exhibited significant PPI in all background sound conditions were unilaterally exposed to loud noise (n = 9).

Guinea pigs were anesthetized with ketamine (Ketaset; 50 mg kg⁻¹, i.p.; Fort Dodge Animal Health Ltd., Southampton, UK) and xylazine (Rompun; 10 mg kg⁻¹, i.p.; Bayer PLC, Newbury, UK), supplemented with further administrations of a mixture of ketamine and xylazine, in a ratio of 15:2 (i.m.), throughout the procedure. A homeothermic heating pad (Harvard Apparatus Ltd., Edenbridge, UK) and a rectal probe were used to monitor core body temperature and maintain it at 38 ± 0.5°C. Auditory brainstem responses (ABRs) were recorded prior to, immediately after, and 8 weeks following AOE to determine hearing thresholds, in the same manner as described previously (8). GPs were placed inside a sound-attenuated booth and remained there for the duration of the ABR recording and AOE. Following the collection of pre-trauma ABR thresholds, GPs were exposed to NBN bursts (duration of 500 ms and inter stimulus interval of 200 ms; center frequency 10 kHz; bandwidth 1 kHz), presented to the left ear only at 120 dB SPL, for 1 h via a 25 mm loud speaker (Peerless DX25, Tymphany, Hong Kong), connected to a 20 mm diameter polyethylene tube in order to form a seal around the ear and maintain a closed sound system. To minimize the risk of damage to the contralateral (right) ear, the right pinna was folded over and a polyethylene tube plugged with cotton wool was placed over the ear. ABRs recorded immediately after AOE confirmed that trauma occurred unilaterally. The AOE protocol was designed to minimize long-term hearing threshold shifts.

We employed a commonly used classification for identifying behavioral evidence of tinnitus [e.g., Ref. (8, 10, 26, 27)]. Baseline gap detection data were pooled across all sessions. These were compared to pooled data acquired 7–8 weeks after AOE. This time-point was selected based on the assumption that tinnitus develops within 7–8 weeks following AOE (6, 13) and that hyperactivity becomes no longer dependent on cochlear input ~6 weeks after AOE (19, 20). PPI at the 7–8 week time-point was expressed as a ratio compared to baseline (before/after). Thus, a value of 1 indicated no change in PPI (relative to baseline), whereas a value <1 indicated a reduction in PPI, while a value >1 suggested that an improvement in PPI. Significant reductions in gap-induced PPI at any background frequency (compared to baseline) were identified using a two-way analysis of variance (ANOVA) with a Bonferroni post hoc test (P < 0.05). GPs that exhibited a significant reduction in gap-induced PPI at one or more background frequencies were categorized as “tinnitus” animals, while those that did not were categorized as “no tinnitus” animals.

Following behavioral confirmation of tinnitus, GPs were anesthetized with urethane (0.5 g kg⁻¹ in 20% solution, i.p.; Sigma, UK), ketamine (40 mg kg⁻¹, i.p.), and xylazine (8 mg kg⁻¹, i.p.), supplemented with further administrations of a mixture of ketamine and xylazine, in a ratio of 15:2 (i.m.), throughout the procedure to maintain areflexia. A single injection of atropine sulfate (0.06 mg kg⁻¹, s.c.) was administered to suppress bronchial secretions. ABRs were then recorded to determine whether any permanent hearing threshold deficits were present. Following ABR recordings, GPs were tracheotomized and respired with 100% oxygen to maintain normal end-tidal CO₂ partial pressure within a range of 28–38 mmHg. Core body temperature was monitored throughout, as described in the previous section. Animals were held in place using a stereotaxic frame, with hollow plastic speculae inserted into the ear canals replacing the ear bars. Throughout the duration of recording, animals remained inside a sound-attenuating chamber. Polyethylene cannulae (>10 cm length, 0.5 mm outer diameter) were inserted into the bullae on each side, in order to equalize pressure across the tympanic membrane while maintaining closed-field stimulation conditions. The posterior fossa was opened to reduce respiratory pulsations of the brain. Craniotomies were performed over the right and left IC (~4 mm diameter) using coordinates described in the guinea pig atlas of Rapisarda and Bacchelli (28). The Dura mater was excised and pairs of electrode arrays were slowly lowered into the brain at 10° to the vertical plane in the medial–lateral orientation. This was not only required to accommodate electrode arrays simultaneously on the two sides but also enabled a perpendicular approach to the isofrequency laminae of the IC (29). The exposed cortex was kept moist with intermittent application of warm 0.9% sodium chloride solution. When necessary, the brain surface was covered in 1.5% agar for stabilization during recording.

To simultaneously record from left and right IC, two arrays of four glass-coated tungsten electrodes (~1–3 MΩ impedance) were attached to separate circuit boards (30). The tips of the electrodes were aligned and separated by ~200 μm. These electrode arrays were connected to a Tucker Davis Technologies (TDT) System 3 via TDT headstage amplifier and a TDT Medusa preamplifier. Electrodes were advanced into left and right IC, and extracellular single-units were recorded (filtered between 600 Hz and 3 kHz). Data were collected using Brainware (software developed by J. Schnupp, University of Oxford, UK).

Etymotic ER-4 earphones (Etymotic Research, Inc., IL, USA) were inserted into the hollow speculae, in order to present auditory stimuli diotically and create sealed acoustic systems. Search stimuli, generated using the TDT System 3, consisted of a wideband noise (duration 50 ms), with cosine-squared ramps lasting 8 ms and a repetition interval of 300 ms. Following online isolation of a single-unit, a frequency-response area was measured by presenting pure tone bursts (50 ms duration; 200 ms repetition interval) over a range of frequencies (50 Hz to ~25 kHz randomly interleaved at 0.25 octave intervals) and sound levels (attenuations of 0–95 dB in 5 dB steps, from a maximum of ~100 dB SPL). This enabled determination of the characteristic frequency (CF) of isolated single-units across multiple electrodes, which could potentially have disparate CFs.

The minimum gap detection threshold (MGDT) was measured for each isolated single-unit in GPs with behaviorally identified tinnitus (n = 7), AOE GPs without tinnitus (n = 2) and in an additional group of control animals (n = 6) that were not noise-exposed. Auditory stimuli comprised a noise/tone burst (duration of 200 ms, on/off ramps of 0.5 ms), followed by a fixed-length period of silence or “gap” (durations of either 1, 2, 4, 8, 10, 20, 50 or 75 ms), and a final noise/tone burst (duration of 50 ms, on/off ramps of 2 ms). Each gap condition was presented in ascending order (20 repetitions, 700 ms repetition interval), for three types of carrier: (i) BBN, (ii) NBN (matched to the behavioral NBN background sound conditions, i.e., 4–6, 8–10, 12–14, or 16–18 kHz), or (iii) pure tones. Pure tones were presented at a frequency that matched the CF of each single-unit, as determined by frequency-response areas. NBN bursts were presented at the frequency range closest to the CF of a given neuron. In all cases, pre-gap and post-gap stimuli were spectrally identical, and were presented at the same sound level.

The sound level was determined based on that used in the behavioral test: as described in our previous work, we determined optimal sound levels of startling stimuli (95, 100, or 105 dB SPL) and background carrier (55, 60, or 70 dB SPL) to maximize baseline PPI for each animal [sound level-dependency test; see Ref. (7)]. In order to better compare behavioral performance and neural MGDTs (determined electrophysiologically), we selected analogous sound levels for both approaches. Although these were not directly comparable owing to a number of methodological differences (e.g., speaker type and anesthesia), this was deemed the most consistent way of selecting an appropriate sound level for MGDT measurement.

Minimum gap detection thresholds of single-units were defined as the minimum gap duration where a significant increase in firing (>2SD above the mean firing rate within the preceding 50 ms) could be detected following the onset of the post-gap stimulus, i.e., a comparison of firing immediately after the gap vs. during or before the gap. These responses were further required to contain a minimum of three spikes (collected over 20 sweeps). MGDTs were assessed using custom-written Matlab software (R2009b, Math-Works, MA, USA). In addition, post-stimulus time histogram (PSTH) plots for each cell were visually inspected for confirmation purposes.

Single-unit MGDTs were calculated for each sound condition: pure tones (at CF), BBN, and NBN. For the NBN condition, single-units were included if the CF of a given single-unit was ≤1 kHz from the lower or upper limits of an NBN frequency range, e.g., single-units with a CF of 3–7 kHz were assessed in response to 4–6 kHz NBN stimuli. This was done to restrict the effect that having a NBN frequency substantially different from the CF may have on MGDTs. A small subset of units exhibited offset responses only, i.e., neuronal firing was suppressed during the presentation of auditory stimuli. For these single-units, MGDTs were determined solely by visual inspection, using a discernible increase in firing following the offset of the gap as an indication that the gap was detected.

The number of responsive neurons to each gap condition was expressed as a percentage of all recorded neurons in tinnitus and control groups. Mean MGDTs were also compared for tinnitus, no tinnitus, and control GPs for each noise condition, and differences between the three experimental groups statistically assessed with a Kruskal–Wallis test with a Dunn’s post hoc test. Data were initially pooled regardless of the side from which they were recorded, as we have previously demonstrated that a unilateral noise exposure resulted in bilateral increases in spontaneous firing rates (8) and this method of analysis preserved larger sample sizes. Analysis of MGDTs across left (ipsilateral) and right (contralateral) IC is further supported by the fact that the behavioral task involves binaural presentation of stimuli. However, MGDT data were subsequently analyzed independently for each side, as the unilateral AOE may have feasibly resulted in differences between sides. Statistical comparisons were made with a two-way ANOVA and Tukey’s HSD post hoc test.

Inferior colliculus single-units were sub-divided according to response profiles exhibited to gap detection stimuli. Three different classes were extracted from these data: (i) if a single-unit responded to the onset of the first 200 ms stimulus, but ceased firing within ~30 ms, it was labeled as an onset response; (ii) if a single-unit showed a response that lasted more than 30 ms of the first 200 ms stimulus, it was designated as a sustained response; (iii) single-units that were predominantly silent throughout the presentation of both the initial 200 ms stimulus and the second 50 ms stimulus, yet responded following the offset of either stimulus, were categorized as exhibiting an offset response. The classifications for onset and sustained-response single-units were based on the scheme of Astl et al. (31), while offset responses were classified in a similar manner to Kasai et al. (32).

The proportion of response types was calculated for control, no tinnitus, and tinnitus groups, and compared. Chi-squared tests were used to determine whether there were any significant differences in the distributions of response types overall or when sub-divided according to side. MGDT data were also sub-divided, according to response profile (and experimental group) and statistically assessed with a two-way ANOVA.

After concluding behavioral testing, single-unit recordings were performed in the left, ipsilateral IC (IC_ipsi) and right, contralateral IC (IC_contra) of control (n = 88 cells), no tinnitus (n = 42 cells), and tinnitus (n = 105 cells) groups of GPs. Figure 2 highlights the three categories under which IC units were classified, and also shows responses to pure tone stimuli with varying gap lengths, as used for determining MGDTs: Figure 2A shows a representative example of a cell responding to the stimulus onset; Figure 2B shows a cell exhibiting a sustained response; Figure 2C shows a cell responding only to the offset of a stimulus. Data from no tinnitus, tinnitus, and control groups were analyzed to determine whether there were any changes in proportions of the types of responses exhibited by IC neurons. The percentage of “onset” and “sustained” single-unit types in response to pure tone stimuli for each group are shown in Figure 3A.

In controls, 26% of cells exhibited onset responses (n = 23), 68% were of a sustained type (n = 60), and 6% displayed offset responses (n = 5). In tinnitus animals, 51% of units were onset-responders (n = 54), while 48% of units were classified as sustained responses (n = 50), and only 1% of units demonstrated offset responses (n = 1). In no tinnitus animals, 62% of units were onset-responders (n = 26), while 38% of units were classified as sustained responses (n = 16). Chi-squared tests were applied to compare the frequency with which different types of units occurred (excluding offset responses due to the extremely low incidence), this revealed that tinnitus GPs – proportionally – exhibited significantly fewer sustained responses than controls and significantly more onset responses [χ²(1) = 29.52, P < 0.0001]. When comparing no tinnitus GPs to controls, there were again significantly fewer sustained responses and significantly more onset responses [χ²(1) = 23.95, P < 0.0001]. There was no significant difference between no tinnitus and tinnitus GPs in response types [χ²(1) = 1.65, P = 0.20]. However, as Figure 3B highlights, overall there were considerably fewer recorded units in no tinnitus GPs (compared to the other two groups) meaning that the sample size was not large enough to determine whether changes in this group were reliable or a result of a sampling bias.

In order to determine whether the change in the proportion of response types between control, no tinnitus, and tinnitus groups was limited to a specific side, data were analyzed for IC_ipsi and IC_contra separately (Figure 3C). Offset units were again excluded from analysis due to their low incidence (n = 6 across all experimental groups). In control animals, there was no significant difference between the two sides: 27% of units recorded from the IC_ipsi exhibited onset responses (n = 15) and 73% exhibited sustained responses (n = 41), compared with 30% onset (n = 8), and 70% sustained (n = 19) from the IC_contra [χ²(1) = 0.10, P = 0.76]. In tinnitus animals, separating data according to side clearly highlighted where response-type differences between the two experimental groups occurred: the proportions of response types recorded from the IC_ipsi were similar to controls; 27% onset responses (n = 8) and 73% sustained responses (n = 22). However, in the IC_contra of tinnitus GPs, 62% of units exhibited onset responses (n = 46), compared with only 38% sustained responses (n = 28). The difference in the proportion of onset vs. sustained response types between IC_ipsi and IC_contra in tinnitus GPs was highly significant [χ²(1) = 46.42, P < 0.0001]. In no tinnitus GPs, the change in the proportion of onset types was evident for both the IC_ipsi and the IC_contra, and there was no significant difference between the two sides: 67% of units recorded from the IC_ipsi exhibited onset responses (n = 22) and 33% exhibited sustained responses (n = 11), compared with 44% onset (n = 4), and 56% sustained (n = 5) from the IC_contra [χ²(1) = 2.07, P = 0.15]. It should be noted, however, that the sample sizes in the IC_contra in this group were very small when separated in this manner.

In summary, AOE resulted in a marked shift in IC single-unit response types, i.e., a significantly higher proportion of units classified as onset-responders, compared with controls. In tinnitus animals, this shift was only found in the IC_contra, which – taken alone – is perhaps not altogether surprising given the unilateral nature of the AOE paradigm. In no tinnitus animals, this shift was evident to a greater degree in IC_ipsi compared with IC_contra, although there was no significant difference between the two sides. The mechanisms underlying the difference in the laterality of effects between no tinnitus and tinnitus GPs are unclear, and are difficult to resolve with the present data.

The mean (±SEM) MGDTs for pure tone stimuli, separated according to experimental group, are shown in Figure 4A. MGDTs were 10.22 ms (±1.75 ms; n = 88) for controls, 14.12 ms (±3.39 ms; n = 42) for no tinnitus GPs, and 13.14 ms (±1.97 ms; n = 105) for tinnitus GPs; no statistically significant differences were found between the three groups (P = 0.09). Pure tone MGDTs were also plotted as a percentage of the total number of cells responding to each gap duration tested in Figure 4D. In control GPs, 99% of single-units exhibited MGDTs of 50 ms or less, while in no tinnitus GPs, MGDTs of 50 ms or less were observed in 90% of cells and in tinnitus GPs, MGDTs of 50 ms or less were observed in 95% of cells recorded.

Figure 4B shows the mean MGDTs, and Figure 4E shows the percentage of gap-detecting cells for each group in response to BBN. Mean (±SEM) MGDTs in response to BBN were 10.95 ms (±1.97 ms; n = 75) for controls, 16.48 ms (±3.58 ms; n = 42) for no tinnitus GPs, and 18.79 ms (±2.27 ms; n = 103) for tinnitus GPs. MGDTs were significantly longer for tinnitus GPs compared with controls (P < 0.05), but not when compared with no tinnitus GPs. However, importantly, the percentage of units with MGDTs of 50 ms or less were similar for control (97%), no tinnitus (92%), and tinnitus GPs (93%). The sample size for responses to BBN and NBN stimuli was smaller than for pure tones, as in some cases cells either did not respond to these stimuli, or exhibited a CF >1 kHz from the NBN frequency range.

Figure 4C shows the mean MGDTs for NBN, while the percentage of gap-detecting cells are shown in Figure 4F. Mean (±SEM) MGDTs were 13.33 ms (±3.64 ms; n = 43) for controls, 20.28 ms (±4.96 ms; n = 25) for no tinnitus GPs, and 19.18 ms (±4.03 ms; n = 33) for tinnitus GPs. No significant differences were evident between the three groups (P = 0.18) and, once again, the percentage of units with MGDTs ≤50 ms were similar for control (95%), no tinnitus (92%), and tinnitus (91%) groups.

To summarize, although MGDTs were significantly increased in tinnitus GPs in response to BBN stimuli, the effects on the percentage of cells capable of detecting gaps of ≤50 ms were negligible, regardless of the characteristics of the stimulus or the experimental group. A lack of dramatic change in neural gap detection in tinnitus GPs is further highlighted by Figures 4G–I, which show the number of units with MGDTs at each gap duration in response to the three different stimulus conditions. For the purposes of clarity, no tinnitus GPs have been omitted from these histograms. These figures clearly demonstrate that many units responded to gaps of short durations, regardless of whether they were recorded from tinnitus GPs or not.

Figures 4J–L show the averaged population responses for each experimental group to gaps with durations of 4, 8, and 50 ms, respectively, using a pure tone carrier. Although the response to the second stimulus onset was weaker in tinnitus GPs for the 4 ms gap condition (Figure 4J) – indicating slightly poorer gap detection – the response to the second stimulus with an 8 ms gap in tinnitus GPs actually slightly exceeded that of control GPs (Figure 4K), and control and tinnitus GPs were equivalent when a 50 ms gap was presented (Figure 4L). This further highlights that, while some minor differences were evident for short-gap durations, the detectability of gaps ≥8 ms was similar for controls and tinnitus GPs. For no tinnitus GPs, responses to 4 ms gaps were clearly reduced compared to both control GPs and tinnitus GPs. For gaps ≥8 ms duration, responses were similar to those of controls. The apparent reduction in sustained firing (following the onset response) in tinnitus and no tinnitus GPs was most likely a result of the proportional reduction in sustained single-unit responses in these groups.

Due to the unilateral nature of the AOE protocol, it was entirely plausible that any changes in MGDTs may have been restricted solely to the IC_contra. Thus, pooling data from both sides may have diluted any AOE-related changes in MGDTs. Consequently, MGDTs in response to pure tones and BBN were sub-divided according to recording side, for control, no tinnitus, and tinnitus GPs (Figures 5 and 6). This was not feasible for NBN data owing to the small sample size.

Figure 5A shows mean MGDTs (±SEM) in response to pure tones for control, no tinnitus, and tinnitus GPs, separated for IC_ipsi and IC_contra. The percentage of gap-detecting cells in IC_ipsi and IC_contra are shown in Figures 5B,C, respectively. Observationally, the IC_contra exhibited shorter MGDTs than the IC_ipsi in controls, no tinnitus, and tinnitus groups, indicating a right-side dominant asymmetry. Statistical analysis revealed a significantly shorter MGDT overall in the IC_contra compared with the IC_ipsi [F_{(1, 229)} = 5.72; P < 0.05], but no significant effect of experimental group [F_{(2, 229)} = 1.07, P = 0.35], and no collicular side × group interaction [F_{(2, 229)} = 0.004, P = 0.99]. Post hoc analyses, comparing sub-divided IC_ipsi control (n = 57), IC_contra control (n = 31), IC_ipsi no tinnitus (n = 33), IC_contra no tinnitus (n = 9), IC_ipsi tinnitus (n = 31), and IC_contra tinnitus (n = 74) recordings, revealed no further differences either as a “within-group” comparison (i.e., IC_ipsi vs. IC_contra) or “between-group” comparison (e.g., control IC_ipsi vs. tinnitus IC_ipsi). The percentage of cells with MGDTs of ≤50 ms were 98% (IC_ipsi) and 100% (IC_contra) for control GPs, while for tinnitus GPs these numbers were slightly lower at 94% (IC_ipsi) and 96% (IC_contra). In no tinnitus GPs, these numbers were lower for the IC_ipsi (88%) and the same as controls for the IC_contra (100%).

Figures 6A–C show mean MGDTs and the percentage of gap-detecting cells in response to a BBN stimulus, separated according to side, for control, no tinnitus, and tinnitus GPs. Overall, there were no significant differences between the two sides [F_{(1, 223)} = 0.30; P = 0.58], nor across experimental groups [F_{(2, 223)} = 1.52; P = 0.22], but a group × side interaction was present [F_{(2, 223)} = 4.43; P < 0.05]. Post hoc analyses indicated that the IC_contra in tinnitus GPs (n = 74) had significantly longer MGDTs than both the IC_ipsi in tinnitus GPs (the “within-group” comparison; n = 34, P < 0.01) and the IC_contra in control GPs (the “between-group” comparison; n = 31, P < 0.01). Thus, neural GDTs to BBN stimuli were significantly worse in IC_contra of tinnitus GPs, which reflects activity generated by the AOE-treated ear. In response to BBN, the percentage of cells with MGDTs of 50 ms or less was 96% (IC_ipsi) and 100% (IC_contra) for control GPs. For tinnitus GPs, the percentage of units with thresholds of ≤50 ms was higher than controls for the IC_ipsi (100%) but lower for the IC_contra (90%). This contrasts with responses to pure tone stimuli, i.e., no change in the trend toward right-side-dominance, simply a small increase in MGDT relative to controls. For no tinnitus GPs, the percentage of units with thresholds of ≤50 ms was 90% (IC_ipsi) and 91% (IC_contra), but the small sample sizes in this group (relative to control and tinnitus GPs) should be noted.

Next, onset and sustained units were compared to establish whether the changes in neural gap detection were more pronounced in either of these sub-classes. Across the three experimental groups (pooled across both recording sides), in response to the pure tone stimulus, onset cells displayed a mean (±SEM) MGDT of 15.74 ms (±2.16 ms; n = 103). The mean MGDT for sustained units was considerably shorter at 7.45 ms (±1.43 ms; n = 126). Offset units were excluded from analysis due to their low incidence in the sample (n = 6 across both groups), although the mean MGDT of offset units was considerably longer at 30.5 ms (±10.81 ms; n = 6), this is consistent with the results of other previous studies (16, 41).

Data were also examined to see whether MGDTs (in response to pure tones, as this was the largest sample available) varied as a function of response type for each experimental group (Figure 7): the mean (±SEM) MGDT in control GPs for onset units was 17.43 ms (±4.19 ms; n = 23), while sustained units had a mean MGDT of 6.75 ms (±1.85 ms; n = 56). In no tinnitus GPs, mean MDGTs were 18.54 ms (±4.55 ms; n = 26) for onset units and 6.94 ms (±4.56 ms; n = 16) for sustained units. Mean MDGTs in tinnitus GPs were 13.62 ms (±2.66 ms; n = 53) for onset units and 8.76 ms (±2.42 ms; n = 50) for sustained units.

Statistical analysis indicated a significant overall effect of response type on MGDT [F_{(1, 218)} = 11.15, P < 0.01]. However, there were no significant differences between experimental groups [F_{(2, 218)} = 0.11, P = 0.90], nor was there an interaction between experimental group and response type [F_{(2, 218)} = 0.61; P = 0.55] or any post hoc differences. In other words, sustained units had significantly shorter MGDTs than onset units, irrespective of experimental group.