PMC

Experimental design and electrode locations. a Auditory–olfactory matching task. Each trial began with a computerized, spoken descriptive word (rose or mint), followed ~5 s later by the presentation of an odor. After smelling the odor, participants verbally indicated whether the odor matched the word. A trial was considered correct if the participant responded "yes" when the odor matched the cue or "no" when the odor did not match the cue. b, c Electrode locations. Red dots on individual subject brains indicate electrodes located in the superior temporal gyrus (b) and piriform cortex (c) for each participant (S1–S7). Blue dots in b show the entirety of implanted parietal grids (S1–S5). S6 and S7 had depth wires implanted with an electrode in superior temporal gyrus. L left hemisphere

Auditory cues induced cross-frequency coupling. a On the left, group-level comodulograms of within-piriform cortex (PC) cross-frequency coupling are shown. Values on the plot represent the modulation index (MI) for each phase amplitude pair (see Methods). On the right, phase–amplitude distributions of the maximal modulatory frequency are shown (corresponding to red and blue circles on the left panel). b Auditory cue-induced changes in phase–amplitude coupling in PC. In the left and middle columns, comodulograms computed from pre-cue ([−5, 0] s) and post-cue ([0, 5] s) time windows are shown separately, with the difference between the two on the far right. c On the left, group-level comodulograms of between auditory (AUD)–piriform cross-frequency coupling are shown. Values on the plot represent the MI for each phase amplitude pair (see Methods). On the right, phase–amplitude distributions of the maximal modulatory frequency are shown (corresponding to red and blue circles on the left panel). d Auditory cue-induced changes in phase–amplitude coupling between AUD and PC. In the left and middle columns, comodulograms computed from pre-cue ([−5, 0] s) and post-cue ([0, 5] s) time windows are shown separately, with the difference between the two on the far right. e, f Phase-amplitude coupling between AUD and PC amplitude is stronger during correct trials. Comodulograms computed separately for correct and incorrect trials for both within-PC coupling (left) and AUD–PC coupling (right) are shown. Increased modulation during correct compared to incorrect trials was only evident for AUD–PC coupling in the theta–beta ranges. In all instances, comodulogram z scores were corrected for multiple comparisons using false discovery rate (p < 0.05, permutation test)

Cue-induced local field potential (LFP) amplitude changes. a Group-level auditory-cue-aligned spectrograms were computed from auditory cortex (AUD) and piriform cortex (PC) LFPs. Areas of statistical significance are outlined in black (false discovery rate (FDR) corrected p < 0.05, permutation test). b Frequency of maximal cue-evoked amplitude modulations in AUD (red) and PC (blue). The average z scores over a time window of [0−1] s following auditory cues are plotted as a function of the frequency. Arrows indicate peak frequencies. The vertical black dotted lines indicate the FDR-corrected threshold for statistical significance in a. c Individual-level analysis of auditory cue-induced responses in PC. Pre-cue and post-cue average low frequency (1–7 Hz) amplitudes are shown for each participant (S1–S7). d Single-trial theta (3–5 Hz) amplitude time-series for each individual participant (S1–S7). On each plot, trials are sorted by latency-to-peak. These plots show raw data that has not been baseline corrected. e Percentage of signal change at peak response frequency (3–5 Hz) in AUD (red) and PC (blue). Arrows indicate the time from auditory cue onset (t = 0) to the peak response. f Peak latency difference between AUD and PC. The histogram (blue bars) indicates the null distribution of permuted differences between AUD and PC latencies. The red line indicates the normal curve fit. The vertical black line represents the actual (non-permuted) peak latency difference (PC−AUD), revealing a statistically significant difference (p < 0.0001, z = 8.81, permutation test)

Auditory cue-induced phase synchronization between auditory and olfactory cortices. a Raw phase-locking value (PLV) computed between auditory cortex (STG) and piriform cortex (PC) is shown for each participant (S1–S2, S4–S7). Black outlined areas indicate statistically significant clusters (false discovery rate (FDR) corrected p < 0.05, Rayleigh test). The dashed line indicates the onset of the auditory cue. Note that we did not compute PLV for S3 because this participant did not complete enough trials for individual-level PLV analysis (see Methods). b Auditory cue-induced PLV increases at the individual participant level. PLV strength before and after auditory cues was compared using a two-tailed paired t test (t(5) = 6.44, p = 0.0013). c Phase synchronization between primary auditory and olfactory cortices was stronger during correct compared to incorrect trials. PLV is shown computed for correct (top) and incorrect (bottom) trials separately. Dashed lines indicate the onset of auditory cues. d Phase synchrony was stronger when the participant's response was correct. Direct statistical comparison between correct and incorrect trials is shown in the upper panel. Black outlines indicate statistically significant clusters (FDR corrected p < 0.05, permutation test) and gray outlines indicate statistically significant clusters at uncorrected p < 0.05. The lower panel shows the averaged PLV time series at the peak frequency. Gray shaded areas denote the standard error obtained through bootstrapping. e PLV strength predicts future accuracy. Bootstrapped PLV values are shown plotted against the mean accuracy of the subset of trials included in each repetition. The strength of PLV for each subset of trials strongly correlates with the future behavioral accuracy (r = 0.943, p = 0.000043, Pearson's correlation). f Auditory cue-induced PLV is maximal between PC and STG. All parietal electrodes for all participants are shown as dots overlaid on the standard Montreal Neurological Institute (MNI) brain. On the left, each dot represents one electrode, color-coded by the strength of cue-induced PLV between PC and that particular electrode; greener colors indicate stronger PLV. On the right, raw PLV values are shown interpolated into a heat map overlaid on the standard MNI brain surface; warmer colors indicate stronger PLV

Control analyses. a Auditory cue-induced responses in piriform cortex (PC) were anatomically specific. Cue-aligned spectrograms for electrodes inside and lateral to PC. Black-outlined clusters indicate statistical significance (false discovery rate corrected p < 0.05, permutation test). Dashed lines indicate the onset of the auditory cue. b Location of PC-bound depth wires for all participants overlaid on the Montreal Neurological Institute standard brain. Each dot represents one electrode along the depth wire for each participant. One patient who had right hemisphere placement of the PC depth wire was mirrored to the left hemisphere. c Spatial distribution of auditory cue responses showing a hot spot in PC. Hotter colors represent larger response magnitudes that were determined by the peak response following auditory cues, computed separately for all electrodes along PC-bound depth wires for all participants. d Responses in depth wires located inside PC were consistently larger than those located outside of PC. Values are shown for each participant (S1–S7) and were compared using a two-tailed paired t test across participants (t(6) = 3.79, p = 0.009). e Auditory cue-induced responses in PC were not driven by respiratory modulations. Auditory cues did not modify respiratory behavior. Raw respiratory signals were aligned to the auditory cue and then averaged, showing no change in breathing following auditory cues (yellow line). Respiratory data aligned to all inhale onsets over the entire experiment (blue line) and the subset of inhale onsets that occurred after the auditory cue and before the odor (red line), show that the cues also did not impact the subsequent breath. Gray shaded areas surrounding each line indicate standard error over participants. The panels on the right show individual participants’ maximal and minimal airflow values during the pre- and post-cue time windows (top), individual participants’ inhale and exhale peaks (bottom left and middle), and individual participants’ inhale volumes for breaths taken before and after the cues (bottom right); n.s. indicates p > 0.05, two-tailed paired t test