the respiratory-related evoked potential (RREP) has been elicited by inspiratory occlusion and graded, easily detected inspiratory resistive (R) loads (10, 14, 15, 17, 28, 35, 44, 51). A relationship between R load magnitude, magnitude estimation, and the short-latency P₁ peak amplitude has been demonstrated (23, 24). There is also a report of a minimum threshold for eliciting the long-latency peaks [N₂, P₂, and P₃₀₀ peaks (3)], but the authors were unable to record the short-latency P₁ and N₁ peaks because they had a limited unilateral scalp electrode montage, used only loads presented at the onset of the inspiration, and were unaware of the existence of the frontal N_f peak (14, 28). Thus there is no report of an inspiratory load threshold for eliciting the primary initial sensory peaks, P₁ and N_f, of the RREP. However, it is known that presentation of no load does not elicit any peaks of the RREP (10, 14, 35, 51). This suggests a threshold for eliciting these early peaks of the RREP. Although it is clear that there is a relationship between the magnitude estimation of inspiratory resistive loads and the early peaks (N_f, P₁, and N₁) of the RREP (24), i.e., load perception and neural measures of cortical neural processing of load information, it is unknown whether the load must exceed the cognitive detection threshold for elicitation of these RREP measures of neural sensory processing. The central hypothesis for this study is, therefore, that the short-latency P₁ and N_f sensory peaks of the RREP can be elicited only by R loads that exceed the R load detection threshold [the added external R load that has a 50% detection probability (ΔR₅₀)].

The RREP cognitive, attention-related peak most commonly studied was the positive peak that occurs ∼300 ms after the application of the load, P₃₀₀ (2, 3, 41, 43–45), and it was present when the subject attended (attempted to detect) to the load. It has been reported that the latency and amplitude of the P₃₀₀ RREP component were correlated with cognitive attention and the stimulus magnitude (2, 3, 42). It has been further reported that the cognitive-related P₃₀₀ peak was not elicited by loads below the detection threshold and no load (3). No-load control RREP recordings have been reported to not elicit the P₁, N_f, or N₁ peaks of the RREP (2, 3, 10, 20, 23, 24, 42). It is likely that there is a R load threshold for eliciting both the short-latency and long-latency peaks of the RREP. If the P₁ and N_f peaks reflect the arrival of the respiratory load-related mechanosensory information and the P₃₀₀ is related to the cognitive cortical processing of afferent information, then it was reasoned that the P₃₀₀ peak would be present only if the P₁ peak was also present. This means that the P₃₀₀ peak should be present only for R loads above the ΔR₅₀ threshold and if the subject attended to the load. We hypothesized that the P₃₀₀ peak would be present when the P₁ peak was also present with loads that exceeded the ΔR₅₀ detection threshold during attend conditions.

Humans can easily detect mechanical loads added to inspiration. The detection threshold for R loads is defined as ΔR₅₀ (45). It has been reported that the probability for R load detection is a function of the background resistance (R_o), i.e., the resistance of the airways plus the resistance of the breathing apparatus (46). The Weber-Fechner law predicts that the detection of a stimulus is a constant fraction of the baseline stimulus intensity (46). For R loads, the Weber fraction (ΔR₅₀/R₀) is ∼0.3 (46). Previous R load detection studies have delivered the loads at the onset the inspiration (8, 26, 46, 48). The standard protocol requires the subject to inspire against the load at the beginning of the inspiratory effort and continue to breathe against the load throughout the inspiration (7, 8, 16, 26, 48, 49). The subject signals detection of the load if the load is of sufficient magnitude for the subject to detect an added load. Zechman and Davenport (48) reported that the detection of the load is related to the breathing pattern with R load detection correlated with the inspiratory airflow (V̇i). Although it is known that R load detection has a Weber fraction of ∼0.3 and that near-threshold load detection is related to the airflow pattern, it is unknown whether R load detection thresholds are affected by applying the load as an interruption of the inspiration. The application of inspiratory loads by interruption is essential for reliably eliciting the RREP (35). This method of load presentation has been used to demonstrate a relationship between magnitude estimation of the load and the amplitude of the primary sensory peak (P₁) amplitude (23, 24). In a previous study of R load magnitude estimation (25), the slopes of log-log relationship between R load magnitude and magnitude estimation were the same when the loads were presented either at the onset of an inspiration or as an interruption of an inspiration. Although the slopes were unaffected by the presentation method, the loads applied as an interruption of inspiration were estimated as smaller, shifting the relationship to the right and increasing the R load intercept (25). This suggests that the detection threshold increased with the presentation of inspiratory loads as an interruption of inspiration. It was essential to demonstrate the relationship between the load presentation methods and the load detection threshold to use the inspiratory-interruption method for determining the relationship between the detection threshold and the RREP. It was hypothesized ΔR₅₀ would increase with a detection paradigm using R loads presented as an interruption of inspiration.

The present study was designed to initially determine the ΔR₅₀ with loads presented at the onset of the inspiration and as an interruption of inspiration. Loads presented as an inspiratory interruption were then used to elicit the RREP. The RREP was recorded using R loads that were above and below the ΔR₅₀ detection threshold determined by interruption of inspiration for each subject. The RREP was recorded under both ignore and attend conditions. The short-latency (P₁, N_f, N₁) and cognitive P₃₀₀ peaks of the RREP were identified, and their amplitudes and latencies were determined for loads above and below the inspiratory interruption ΔR₅₀.

METHODS

The protocol for this experiment was reviewed and approved by the Institutional Review Board of the University of Florida.

Subjects

Twelve healthy nonsmoking subjects were tested (5 men and 7 women). The mean age of subjects was 24.6 ± 4.5 yr (19–33 yr). Mean height and weight were 171.3 ± 12.8 cm and 68.7 ± 13.2 kg, respectively. The subjects were in good general health based on self-reported medical history of no chronic or acute neurological or respiratory disease. The subjects refrained from consuming alcohol and caffeinated beverages for at least 24 h before the study. Each subject was brought to the laboratory and informed of the nature of the study. The general nature of the experiment was described, and each subject's consent was obtained. The subjects participated in two experimental protocols on a single day. Protocol 1 consisted of the detection of R loads using two different load presentation methods: 1) R loads presented at the onset of inspiration and 2) R loads presented as an interruption of inspiration. Protocol 2 consisted of the recording of the RREP using R load magnitudes that were above and below the ΔR₅₀ threshold. The loads used in Protocol 2 were determined from the R detection trial with loads presented as an interruption of inspiration. Protocol 1 preceded Protocol 2 for all subjects.

Protocol 1: R Load Detection Threshold

Subject preparation.

Pulmonary function was measured. The total respiratory resistance was measured with the impulse oscillometry method (Jaeger MS-IOS, Sensormedics). Forced vital capacity (FVC) and forced expiratory volume in 1 s (FEV₁) were recorded for all subjects with same apparatus. Three FVC measurements were performed on each subject, and the highest value was used for analysis. A FEV₁ above 70% of predicted value was required to admit into the study. All subjects met this inclusion criterion. Pulmonary function was within the normal limits for all subjects. The results of pulmonary function tests are shown in Table 1.

The subject was then seated in a chair with the head, neck, and back supported. The subject respired through a mouthpiece and nonrebreathing valve (2600 series, Hans Rudolph). Care was taken to suspend the valve to eliminate the need for the subject to contract mouth muscles yet maintain an airtight seal. The subject wore a nose clip. The R loads were sintered bronze disks placed in series in a Plexiglas tube (loading manifold), with stoppered ports between disks. The loading manifold was connected to a pneumotachograph, which was connected by reinforced tubing to the inspiratory port of the nonrebreathing valve. Resistances were selected by removing the stopper and allowing the subject to inspire through the selected port. The background resistance of the system (R_app) was 1.85 cmH₂O·l⁻¹·s. The loading manifold was hidden from the subject's view. Mouth pressure (Pm) was recorded from a port in the center of the non-rebreathing valve using a differential pressure transducer and a signal conditioner. V̇i was recorded using a differential pressure transducer connected to the pneumotachograph. V̇i was integrated to obtain the inspired volume (Vi). Pm, V̇i, and Vi were digitalized at 200 Hz (PowerLab/8s, ADInstruments, Mountain View, CA) and stored on the computer for later analysis. The subjects were monitored with a video camera through the study.

Protocol.

The subject was seated in a sound isolated chamber, separated from the investigator and experimental apparatus. Similar to our laboratory's previous R load detection studies (16, 48, 49), a series of suprathreshold and near-threshold R loads were presented in practice session to familiarize the subject with the load sensation and the range of the loads. The subjects were instructed to press the signal button if they detected a load. The signal button was held in their dominant hand and pressed with their thumb.

During the experimental trials, the subjects listened to the music of their choice, which masked experimental sounds. Two methods of presentation were used: 1) R loads presented at the onset of inspiration and 2) R loads presented as an interruption of inspiration. With inspiratory interruption, the R loads were applied after the onset of inspiratory phase, at the beginning of rising phase of V̇i as indicated by the analog display of the airflow signal on the oscilloscope. Onset R load presentations were applied by selecting the load during expiration resulting in the following inspiration against the load beginning at the onset of inspiration. The sequence of load presentation methods was randomized. Eight R load magnitudes (0.20, 0.80, 1.24, 1.64, 2.48, 3.26, 6.95, and 11.46 cmH₂O·l⁻¹·s) were used. Each R load was presented 10 times in a randomized block order, with each individual load presentation separated by 3–6 control breaths. The detection signal was recorded on the computer polygraph system indicating detection of the load. Thus there were 2 experimental trials with 80 loads presented in each trial. The trials differed in the method of load presentation. The subject was allowed a minimum of 5-min rest period off the breathing apparatus between the two trials. The total duration of the Protocol 1 experiment, including preparation time, was ∼1 h.

Data analysis.

All experimental data are presented as means ± SD. The detection threshold was determined separately for each experimental trial. The number of detected loads was summed for each load magnitude. The number of detections was divided by the number of presentations, and this percent detection was plotted against R load magnitude (ΔR). The detection threshold was defined as the R load magnitude corresponding to 50% detection (ΔR₅₀). The ΔR₅₀ was then divided by the total background resistance (R_o), which was the sum of the resistance of the breathing apparatus (R_app) plus the subject's total airway resistance. The resultant ΔR₅₀/R_o was determined for each subject. The ΔR₅₀ and the ΔR₅₀/R_o were also measured for each load presentation method. Comparison of ΔR₅₀ and the ΔR₅₀/R_o between the two trials was performed by using a two-tailed paired t-test. Pearson correlation was performed to test the correlation between two threshold measurements. The significance criterion was set at P < 0.05.

The peak Pm for each R load magnitude was measured from the computer polygraph recordings. The peak Pm for the two load presentation methods was analyzed with Tukey's honestly significant difference test. The significance criterion was set at P < 0.05.

Protocol 2: RREP

Subject preparation.

Protocol 2 was initiated after the completion of the detection study. The methods used in this part of the study were similar to previous RREP recordings with graded R loads (23). The subject was allowed at least a 15-min rest after finishing Protocol 1. The subject was tested in a separate room configured for RREP recordings. Tin electrodes were placed on both ears. The circumference of the head was measured, and an appropriately sized electrode cap with integral tin electrodes was placed on the head. EEG activity was recorded from 12 scalp sites: F₃, F_Z, F₄, C₃, C_Z, C₄, C₃′, C_Z′, C₄′, P₃, P_Z, and P₄ based on the International 10–20 System. EEG activity was referenced to the joined ear lobes. Vertical electrooculogram (EOG) activity was monitored with bipolar electrodes placed lateral to the canthus of the right eye. The impedance level for each electrode is checked and maintained below 5 kΩ. The EEG activity was monitored with an oscilloscope monitor. The electrode cap was connected to an electroencephalograph system (Neurodata 12, Teledyne). The EEG activity was band-pass filtered at 0.3 Hz to 1 kHz, amplified 50 k, led into an online signal averaging computer system, and digitized at 2.5 kHz (model 1401, Cambridge Electronics Design). The Pm signal was recorded by a differential-pressure transducer (model MP-45, Validyne Engineering), digitized, and led into the online signal-averaging computer system. The Pm was also led to an oscilloscope and monitored by the investigator. Pm was monitored continuously and used to time the presentation of the R loads.

A mouthpiece and non-rebreathing valve were connected to a breathing apparatus similar to that used for the detection trials. The inspiratory port of the non-rebreathing valve was connected to the loading manifold by reinforced tubing. A custom-designed gas-activated occlusion valve was connected to the first port of the loading manifold. Activation of the occlusion valve shunted V̇i through the selected R load. Three R loads were used for testing that spanned the detection threshold. The R₁ load (0.80 cmH₂O·l⁻¹·s) was below the ΔR₅₀ for all but one subject. This subject had a ΔR₅₀ of 0.60 cmH₂O·l⁻¹·s, and the R₁ load for this subject was 0.20 cmH₂O·l⁻¹·s. The R₁ loads for all subjects were ∼30% of the ΔR₅₀. The R₂ load (6.95 cmH₂O·l⁻¹·s) was above the ΔR₅₀, and >50% of the R₂ load presentations were detected by all subjects in the inspiratory-interruption detection trials. The R₃ load (11.46 cmH₂O·l⁻¹·s) was easily detected with 100% detection for inspiratory-interruption presentations in all subjects. In addition, R_o was presented. The subject wore a nose clip when respiring on the breathing apparatus. The loading manifold was hidden from the subject's view. Pm was recorded from a port in the center of the non-rebreathing valve. During the experiment, the inspiratory load was presented by silently inflating the occlusion valve after the onset of the inspiration, after the onset of the rising phase of inspiratory airflow as indicated by the analog display of the airflow signal on the oscilloscope. Each loaded breath was separated by three to six unloaded breaths. A transistor-transistor logic pulse generated by the inspiratory occlusion valve controller triggered the collection of 50 ms of pretrigger and 950 ms of posttrigger EEG and Pm data. The duration of the load presentation was ∼500 ms.

Protocol.

Subjects were seated comfortably in a reclining lounge chair with their back, neck, and head supported. There were two conditions: ignore and attend presented with a protocol similar to our laboratory's previous studies (51). For each condition, the four inspiratory load levels (R_o, R₁, R₂, and R₃) were presented in an order to minimize temporal, order, and sequence effects as described previously (23, 24). Subjects received 1 trial of 200 presentations for each condition, with a rest period between each trial. Within each trial, load presentations were tested using 40 blocks of 5 presentations each, with the same load presented within a block to minimize manifold manipulations. Ordering of the different blocks was randomly determined with the restriction that each successive set of four blocks contained only one block of each load level. Thus, over the trial of 200 presentations, each block of 5 presentations of a given load was presented 10 times for 50 presentations of each load magnitude. There was 1 trial for each condition, making 50 presentations of each load magnitude (and no load) available for signal averaging.

The order of the two conditions, ignore and attend, was randomized. During the ignore condition, the subject watched a videotape movie. During the attend condition, the subject listened to music and signaled detection of a load similar to Protocol 1. With the attend condition, the subject was asked to signal the detection of the R load to ensure that attention to sensing the load was maintained. However, because this task was designed to maintain the attention rather than assess the detection performance, these detection data were not analyzed and were discarded. Including subject preparation time, the entire study session for Protocol 2 lasted ∼3 h.

Data analysis.

For each load presentation, a 1,000-ms epoch of EEG activity and Pm was stored on disk for subsequent computer signal averaging (Signal 2, Cambridge Electronic Design). Each load magnitude was averaged separately. The computer stored the individual load trials in the order of presentation. The averaged response for an individual load magnitude was obtained as described previously (23). An individual presentation was included in the average if there was 1) a stable prestimulus EEG activity baseline, 2) no EEG transient activity exceeding ±50 μV, 3) no EOG eye blink, and 4) the onset of the Pm change related to the load aligned with previous presentations. This analysis was repeated for each load magnitude and the no-load control. There were no peaks in the no-load control average that corresponded to the peaks observed with the load-elicited RREP. The EEG activity was averaged independently for each load magnitude for each condition. The presence, latency, and amplitude of RREP components (P₁, N_f, N₁, and P₃₀₀) were determined from the averaged EEG traces. Based on previous peak localization studies (14, 28, 41, 42), the N_f peak was analyzed in the F₃ and F₄ electrodes, the P₁ peak was analyzed from the C₃′ and C₄′ electrodes, the N₁ peak was analyzed from the C_Z electrode, and the P₃₀₀ peak was analyzed from the C_Z and P_Z electrodes. Peak latencies were measured as the time from the onset of the occlusion (indicated by the change in Pm) to the EEG peak. The zero-to-peak amplitude was recorded at the peak of each component. The nomenclature for the peaks is based on previous reports (10, 14, 23, 24, 28, 41–43). The RREP components were identified as follows: N_f was the negative peak in the frontal region occurring within a latency range of 25–45 ms; P₁ was the positive peak in the central region occurring within a latency range of 45–65 ms; N₁ was the negative peak in the vertex (C_Z) occurring within a latency range of 85–125 ms; and P₃₀₀ was the positive peak in the C_Z and P_Z scalp locations occurring within a latency range of 250–350 ms. Averaged RREPs were obtained for each load magnitude, each condition, and each subject.

The descriptive statistics of all the variables were calculated and are expressed as means ± SD. The differences in latencies and amplitude of short-latency RREP components were compared by using repeated-measures ANOVA and Tukey's honestly significant difference test to determine the effects of load magnitude and condition. Because no P₃₀₀ peak was present under ignore condition, the repeated-measures ANOVA was performed to determine the difference in amplitude and latency between different load magnitudes. Significance level was set at P < 0.05.

RESULTS

V̇i decreased, whereas Pm became more negative, compared with control breathing when the R loads were delivered before the onset of inspiration. The pressure became more negative as the load magnitude increased (Fig. 1). There were no significant differences in the peak Pm between onset and interruption load presentations. The mean detection threshold (Fig. 2) for the onset-inspiration method was 2.64 ± 0.89 cmH₂O·l⁻¹·s (range = 1.24–4.35 cmH₂O·l⁻¹·s). The mean ΔR₅₀/R_o for loads presented at the onset of the breath was 0.45 ± 0.13. For inspiratory-interruption delivered loads, V̇i decreased sharply after application of the load, and the Pm had a corresponding rapid decrease. The magnitude of the pressure change increased with increasing load magnitude (Fig. 1). The mean detection threshold for inspiratory interruption presentations (Fig. 2) was 2.66 ± 1.22 cmH₂O·l⁻¹·s (range = 0.60–5.10 cmH₂O·l⁻¹·s). The mean ΔR₅₀/R_o for loads presented interruption of the breath was 0.44 ± 0.18. There was no significant difference in detection threshold between two load presentation methods (P > 0.05). A significant correlation was found in R load detection threshold between the two load presentation methods (r = 0.67, P < 0.05). There was a significant difference for percent detection among the different load magnitudes for both presentation methods (P < 0.001); as the load magnitude increased, the percent detection increased (Fig. 2). However, no significant interaction was found between presentation method and load magnitude percent detection.

R Load RREP Threshold

The RREP was elicited with the R₂ and R₃ loads, which were above the ΔR₅₀ of all subjects, during both ignore and attend trials. The R₁ and no-load R_o did not elicit RREP peaks in any of the subjects (Fig. 3).

RREP ignore condition.

The R₂ and R₃ loads elicited the N_f peak recorded in the F₃ and F₄ electrodes (Fig. 3). The mean N_f amplitudes for R₂ and R₃ are summarized in Table 2. The N_f amplitudes with the R₃ load for F₃ and F₄ were significantly greater (P < 0.01) than for R₂ (Fig. 4). There was no significant difference in the N_f latencies (Table 3) between R₂ and R₃ (Fig. 5). The P₁ peak was elicited by the R₂ and R₃ loads and recorded in the C₃′ and C₄′ electrodes (Fig. 6). The mean P₁ amplitudes are summarized in Table 2. The P₁ amplitudes (Fig. 4) with the R₃ load for C₃′ and C₄′ were significantly greater (P < 0.05) than for R₂. There was no significant difference in the P₁ latencies (Table 3) between R₂ and R₃ (Fig. 5). The R₂ and R₃ loads elicited the N₁ peak recorded in the C_Z electrode (Fig. 6). The mean N₁ amplitudes (Table 2) were not significantly different between R₂ and R₃ (Fig. 4). There was also no significant difference in the N₁ latencies (Table 3) between R₂ and R₃ (Fig. 5). The P₃₀₀ peak was not observed with any of the R loads in the ignore condition.

RREP attend condition.

Attention to the load resulted in a change in the RREP waveform with the presence of a large amplitude P₃₀₀ peak in the attend condition (Fig. 7). The R₂ and R₃ loads elicited the N_f peak recorded in the F₃ and F₄ electrodes. The mean N_f amplitudes for R₂ and R₃ are summarized in Table 2. The N_f amplitudes with the R₃ load for F₃ and F₄ were significantly greater (P < 0.05) than for R₂ (Fig. 8) There was no, however, significant difference for the N_f amplitudes between the attend and ignore conditions (Table 2). There was also no significant difference in the N_f latencies (Table 3) between R₂ and R₃ (Fig. 9) and no significant difference for the N_f latencies between the attend and ignore conditions (Table 3). The R₂ and R₃ loads elicited the P₁ peak recorded in the C₃′ and C₄′ electrodes. The P₁ amplitudes (Table 2) with the R₃ load for C₃′ and C₄′ were significantly greater (P < 0.05) than for R₂ (Fig. 8). There was also no significant difference for the P₁ amplitudes between attend and ignore conditions (Table 2). There was no significant difference in the P₁ latencies (Table 3) between R₂ and R₃ (Fig. 9) and no significant difference for the P₁ latencies between attend and ignore conditions (Table 3). The N₁ peak was elicited by the R₂ and R₃ loads, recorded in the C_Z electrode. The N₁ amplitude (Table 2) with the R₃ load for C_Z was significantly greater (P < 0.01) than for R₂ (Fig. 8). The N₁ amplitude for the R₂ load in the attend condition was not significantly different from the N₁ amplitude for the R₂ load in the ignore condition; however, the N₁ amplitude (Table 2) was significantly (P < 0.01) greater for R₃ in the attend condition than for the same load in the ignore condition (Fig. 8). There was no significant difference in the N₁ latencies (Table 3) between R₂ and R₃ (Fig. 9) and no significant difference for the N₁ latencies between attend and ignore conditions.

The P₃₀₀ was observed with R₂ and R₃ loads only in the attend condition (Fig. 10). The R₂ load elicited the P₃₀₀ peak recorded in the P_Z electrode. The P₃₀₀ amplitudes (Table 2) with the R₃ load for P_Z was significantly greater (P < 0.01) than for R₂ (Fig. 8). There was no significant difference in the P₃₀₀ latencies (Table 3) between R₂ and R₃ (Fig. 9).

DISCUSSION

The results of this study demonstrate an R load threshold for eliciting the short- and long-latency peaks of the RREP. The RREP is only elicited with R loads that exceed the detection threshold. This study further demonstrated that the detection threshold and Weber ratio for R loads presented at the onset of the breath or as an interruption of inspiration are the same and rejects the hypothesis that the increase in the magnitude estimation intercept with inspiratory interruption loads reflects an increase in the detection threshold. This study supports the hypothesis that there is mechanical threshold for respiratory afferents to elicit cognitive awareness of an R load and that this cognitive threshold must be exceeded for the RREP neural measure of cerebral cortical activity to be present. The results further suggest that cognitive awareness of inspiratory loads is mediated by a central neural threshold gating process.

R Load Detection

R load detection is airflow dependent with the detection of near-threshold loads occurring at the point of peak airflow (26, 48). The generation of the inspiratory driving pressure creates the pressure gradient for air to flow into the lung. R loads oppose the movement of air and result in a reduction in airflow (30, 47). Pleural pressure becomes more negative, and the transdiaphragmatic pressure (Pdi) is augmented (49). The pattern of these mechanical changes is directly related to the airflow and the load magnitude. Resistive load magnitudes near the threshold of detection, presented at the onset of the breath (48), produce very small changes in airflow (from control), but they are associated with a more negative Pm and augmented Pdi (49). The Pdi becomes augmented above control progressively earlier in the inspiration as the load magnitude increases. This is also associated with a progressive increase in the detectability of the loads and decrease in the detection latency. This study (49) demonstrated that the detection of an R load was primarily dependent on the change in Pdi. There was a Pdi threshold which, when exceeded, resulted in signaling detection of the load at a constant latency. Thus the change in latency with the change in load magnitude was due to differences in the time required for Pdi to exceed the mechanical threshold in addition to a relatively constant central neural processing time for signaling detection. If it is the change in Pdi that is a major determinant of the detection of the load, then this may explain why the detection threshold for onset and interruption load presentations were the same. A specific load magnitude at a constant respiratory drive will generate the same change in Pdi. The load interrupting the inspiratory effort decreases the airflow and augments inspiratory Pm as a function of load magnitude not timing of the load presentation. The inspiratory interruption loads thus produce a similar magnitude of change in mechanical impediment to inspiration, resulting in the same change in Pm and Pdi, resulting in a similar detection threshold.

The detection of inspiratory loads is dependent on the background mechanical status of the subject (46). The 50% detection threshold for R loads is a constant ratio of the background load: ΔR₅₀/R₀ = k (where k is a constant) (46). The results of this study are consistent with previous reports of this relationship, and the ΔR₅₀/R₀ is not altered by the load presented at the onset of the breath or as an interruption of inspiration. In a previous study, however, R loads presented by inspiratory interruption reduced magnitude estimation of R load compared with the same loads presented at the onset of inspiration (25). This difference was a change in the magnitude estimation of all loads, yet the slope of log-log relationship between magnitude estimation and load magnitude was the same (25). This suggested that, whereas the absolute magnitude of the load was decreased with inspiratory interruption, the sensitivity of the subject to the load did not depend on load presentation method (25). This also means that the effect of inspiratory load pattern is different for detection and magnitude estimation of the load. Of importance to the present study, the R threshold for eliciting the RREP is not a function of the presentation method but due to the load detection threshold.

RREP Detection Threshold

The RREP has been reported to be elicited by inspiratory occlusion and R loads (2, 3, 10, 14, 17, 22–24, 28, 35, 37, 41–43, 51). The short-latency components P₁, N_f and N₁ reflect the activation of cortical neurons by afferent signals ascending somatosensory pathways to the sensory-motor cortex (11, 13, 17, 28, 50) and the supplementary motor area (28). These peaks of the RREP are determined mainly by the physical characteristics of the stimuli. In the present study, the RREP was only present with the presentation of suprathreshold R loads similar to the effect of mechanical tactile stimulation somatosensory-evoked potentials (36). The correlation between the load magnitude necessary to elicit detection and the load magnitude necessary to elicit the peaks of the RREP suggests that these peaks are neural indicators of cortical sensory processing of inspiratory mechanical information.

The first peak of the RREP, P₁, was a positive voltage that was due to the dipole that occurs when a cerebral cortical column was depolarized by the arrival of activity from a population of afferents that were activated by the occlusion stimulus (10, 14, 17, 28). This first peak signals the arrival of the afferent information at the somatosensory cortex. Previous studies have confirmed that the amplitude of P₁ peak is significantly correlated with R load magnitude and magnitude estimation (24, 42). This was confirmed in the present study with the amplitude of the P₁ peak greater for R₃ than the R₂ load. The detection threshold for the P₁ peak has not been previously reported. Bloch-Salisbury et al. (3) were unable to identify the P₁ peak in their study of the R load threshold for the RREP. They used unilateral recording sites referenced to the mastoid processes. The N₁ and P₃₀₀ peaks can be reliably recorded with their electrode montage and reference, but it is very difficult to observe the P₁ at their F₃′, C_Z, and P₃′. In addition, they used only onset load presentations, which elicit smaller amplitude RREP peaks for P₁ and N₁. Onset load presentations have variable peak latencies that depend on the subject's respiratory drive (10). The peak latency should increase with decreasing load magnitude (48), making is very difficult for identifying and analyzing the onset P₁ peak (3). Thus the present study extends their study (3) by using a recording electrode montage that will reliably express the P₁ peak and a load presentation method (inspiratory interruption) that controls for the time required for the load to exceed the mechanoreceptor threshold. The results from this study also demonstrate that once the threshold for eliciting the P₁ peak is exceeded, similar to our laboratory's previous results (23, 24), there is a direct relationship between the magnitude of the load and P₁ peak amplitude, R₃ was greater that R₂. Thus, the first stage of neural processing is eliciting mechanosensory information of sufficient magnitude to activate somatosensory cortical neurons. When the loads exceeded the detection threshold, the P₁ peak amplitude then correlates with the load magnitude and magnitude estimation of the load (23, 24).

The N_f peak was found in the frontal region (14, 28). The N_f peak was not reported with cephalic referenced RREPs (10, 35) nor with electrodes referenced to the mastoids (3). Logie et al. (28) found the source localization for the frontal N_f appears to be elicited by a sensory projection to the lateral premotor cortices. The N_f peak has received little specific analysis. It has been reported that the N_f peak is the result of a neural pathway that is a parallel activation to the somatosensory P₁ activation (14). The N_f peak was elicited only with loads that exceeded the detection threshold. When the N_f peak was found with suprathreshold loads, the amplitude of the N_f peak increased with increasing load magnitude. This suggests that the N_f peak amplitude, similar to the P₁ peak, is a function of the stimulus intensity. The role of this frontal cortical activation in R load detection and information processing, however, remains unknown.

The relationship between R load magnitude, N₁ peak threshold, and N₁ peak amplitude relationship to load magnitude has not been reported. A previous RREP threshold study (3) was unable to analyze the N₁ peak because they could not reliably record the N₁ peak using onset load presentations. In the present study, we demonstrate that the N₁ peak was present with loads that exceed the detection threshold. In the absence of attention to the load (ignore condition), however, the amplitude of the N₁ peak did not vary with suprathreshold load magnitudes. Localization studies have shown the greatest N₁ amplitude at the vertex, C_Z (40, 42). Harver et al. (22) reported that N₁ was unaffected by attention in young subjects. However, Webster and Colrain (42) reported that the N₁ peak amplitude increased with attention to inspiratory occlusions. These latter investigators suggested that the N₁ peak is a compound peak consisting of two negative components, one unassociated with attention and a second attention related (41, 42). This is supported by reports that the auditory N₁₀₀ peak is a composite of two negative components, one related to attention and the other insensitive to attention (21, 31–33, 39). The present study is consistent with a dual source for the N₁ peak. There was a significant increase in N₁ amplitude for the R₃ load with subjects attending to the load. The R₂ attend N₁ amplitude was nonsignificantly greater than for R₂ with the ignore condition. Thus the relationship between load magnitude and stimulus magnitude for the ignore condition did not reach significance primarily because of a smaller R₃ elicited N₁ peak in the ignore condition. When the subject attended to the loads, the N₁ peak amplitude increased, and a significant load magnitude relationship was evident. These results suggest that the RREP N₁ peak is a function of the stimulus parameters and is modulated by subject attention to the stimulus. These results are also consistent with the suggestion (32, 33) that the N₁₀₀ peak is indicative of a triggering process related to the perception of a nonspecific stimulus.

The P₃₀₀ peak is dependent on the subject's attention to the stimulus (41). The present study demonstrated that the P₃₀₀ peak is absent with undetectable loads and if the P₁ peak is not present. When the P₃₀₀ peak was present with suprathreshold loads, there was a direct relationship between R load magnitude and P₃₀₀ amplitude; R₃ P₃₀₀ amplitude was greater than for R₂. These results are consistent with previous reports on the relationship between RREP P₃₀₀ amplitude and R load magnitude (2, 3, 42). The P₃₀₀ displayed a topographic distribution (9, 44) showing centroparietal dominance identical (29) to that seen in the same subjects in response to target stimuli in an auditory oddball task suggesting that the RREP P₃₀₀ is similar in origin to the auditory P₃₀₀. The present results support the load magnitude effect on the amplitude of RREP P₃₀₀ peak. The R₃ load elicited significantly greater P₃₀₀ peak amplitude compared with the R₂ load. The P₃₀₀ was not present with loads below the detection threshold despite the fact that the subjects were attending to the experimental trial. Thus the RREP P₃₀₀ peak is observed only when the R load is of sufficient magnitude to elicit somatosensory cortical activity and the subject attends to detecting the load. This result suggested that the presence of P₃₀₀ peak is correlated with the presence of short-latency RREP components.

This relationship between short-latency RREP components and long-latency RREP components would be expected because respiratory mechanoreceptor information arrives at cerebral cortex, eliciting short-latency stimulus dependent neural activity. This initial cortical activation is then followed by attention related neural activity. The N₁ peak may represent the triggering or gating process related to subjects attending to the stimulus, which is then followed by cortical and subcortical neural activity related to cognitive processing of the load information (P₃₀₀). The pattern of RREP peaks, therefore, represents neural indicators of the temporal sequence of respiratory sensation.

Eupneic breathing normally is not consciously perceived. This suggests that during normal ventilation, respiratory sensory information is gated-out of cognitive centers. When ventilation is obstructed, stimulated, challenged, or attended to, cognitive awareness of breathing occurs (7, 12, 16, 34, 38, 46). The perception of an intrinsic change in respiratory mechanics is an essential component of the cognitive awareness of the onset of an obstructive event. The relationship between the stimulus threshold and central neural activation elicited by this respiratory mechanical stimulation suggests a gated process. Sensory gating has been demonstrated in auditory, visual and somatosensory modalities (1, 5, 6, 19, 27). Modality-specific activation of cortical neural processing centers depends on a change in neural activity that gates-in modality specific information to the brain information processing centers (1, 4–6, 18, 19, 27). This activation leads to cognitive awareness of the modality. The significance of gating-in and gating-out sensory modalities is the need to attend to essential interoceptive physiological functions such as breathing. Respiratory afferent input is normally gated-out from cognitive brain systems, and individuals breathe without conscious attention to their ventilation. Changing the status of the respiratory system alters the input of this information. Respiratory information is then gated-in, eliciting a cognitive awareness of breathing. The transition between breathing that does not reach consciousness to cognitive awareness of breathing suggests neural information processing that gates-in respiratory sensory information. The neural processing of this respiratory information is reflected by the temporal and spatial brain activities that are the foundation of the peaks of the RREP. Hence, the present study is consistent with respiratory cognitive awareness of breathing as a gated process. The absence of the RREP with loads below the detection threshold and the presence of the RREP elicited by suprathreshold loads are consistent with the gating of these neural measures of respiratory mechanosensory information processing.

GRANTS

This study was supported by National Heart, Lung, and Blood Institute Grant HL-48792.

FOOTNOTES

• The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.