Article | Open

# Effect of adapter duration on repetition suppression in inferior temporal cortex

• Scientific Reportsvolume 7, Article number: 3162 (2017)
• doi:10.1038/s41598-017-03172-3
Accepted:
Published:

## Introduction

Most inferior temporal (IT) neurons decrease their response when repeating a visual stimulus1,2,3,4,5,6,7,8,9,10. The mechanisms underlying this repetition suppression are still poorly understood11. The proposed mechanisms range from firing-rate dependent fatigue and synaptic depression to top-down driven expectation suppression. So far, no evidence for top-down expectation-induced repetition suppression has been obtained in nonhuman primates10, 12. Hence, the most parsimonious explanation of repetition suppression in macaque IT currently relies on bottom-up or local fatigue mechanisms, such as firing rate dependent fatigue or reduced input due to synaptic depression or adapted afferents6, 7, 11.

## Methods

### Subjects

Two rhesus macaques (Macaca mulatta; male monkey D and female monkey K) served as subjects. Animal care and experiments were carried out in accordance with the national and European guidelines and were approved by the Animal Ethics Committee of the KU Leuven.

Details about implants and surgery can be found in Kaliukhovich and Vogels10 and will only be briefly summarized here. The placement of the plastic recording chamber was guided with a preoperative magnetic resonance imaging (MRI) scan and verified with MRI scans obtained postoperatively before and in-between recording sessions. For the latter MRI scans, the recording chamber, with the Crist grid located at the same position as during the recordings, was filled with a 1% solution of the gadolinium-based contrast agent Gadoteric acid (Dotarem). This combined with the insertion of tungsten wires, enclosed in glass capillaries, in the grid at 5 locations enabled visualization of the recording chamber, grid and electrode trajectories. Recording positions were estimated based on the MRI visualization of these markers combined with the microdrive depth readings of the white/gray matter transitions relative to the grid base.

Recordings were made from the lower bank of the rostral superior temporal sulcus of the right hemisphere. The anterior-posterior coordinates of the estimated recording positions ranged between 9 and 12 mm and between 15 and 16 mm anterior to the auditory meatus in monkeys D and K, respectively. The corresponding medial-lateral coordinates in those monkeys ranged between 24 and 25 mm, and 20 and 23 mm lateral to the midline.

### Recordings

Well-isolated extracellular action potentials of single IT neurons were recorded with tungsten microelectrodes. The electrode was lowered with a Narishige microdrive through a guide tube that was fixed in a Crist grid. The guide tube was grounded and served as a reference. Amplification and filtering was performed by a Plexon data acquisition system (Plexon Inc.). Recorded signals were preamplified with a headstage having an input impedance of >1GΩ. The signal was bandpass filtered between 250 to 8000 Hz for spikes and 0.7 and 170 Hz for LFPs. Action potentials from single cells were isolated online using the ‘time-window discrimination’ tool provided by the Plexon data acquisition system. Triggered spike-wave forms were saved at 40 kHz for later offline analysis (Offline Sorter; Plexon) in which single unit isolation was re-checked. Stable single unit recordings are expected to be biased towards large pyramidal cells. To overcome this bias, we recorded in some sessions also multi-unit activity by employing a low discrimination threshold level. Note that multi-unit activity may overrepresent the activity of interneurons since the latter have higher firing rates than pyramidal neurons and thus will contribute more to the multi-unit signal compared to a single unit signal. LFPs were recorded simultaneously with the same electrodes.

Eye position was measured online with an infrared-based eye tracking system (ISCAN EC-240A, ISCAN Inc.; 120 Hz sampling rate). The analog eye movement signal was saved with a sampling frequency of 1 kHz. In all tasks, we employed fixation windows that measured maximally 2° on a side. Eye positions, stimulus and behavioral events were recorded simultaneously with spiking activity and stored for later off-line analysis on a computer that was synchronized with the Plexon data acquisition system.

### Stimuli

We employed a stimulus set identical to that used in some of our previous studies on adaptation in macaque IT7, 8. The stimulus set consisted of 52 color images of 26 object classes (2 images per class) including human and monkey faces, human and monkey bodies, mammals, birds, fish, snakes, insects, trees, fruits, fractals and manmade objects. The size of the stimuli (maximum of horizontal and vertical dimensions of the bounding box) was approximately 5° of visual angle. The stimuli were presented on a uniform gray background on a CRT monitor (Phillips Brilliance 202P4, frame rate = 60 Hz, resolution = 1024 × 768 pixels) located 60 cm from the subject’s eyes.

### Search test

While advancing the microelectrode in IT, we searched for responsive neurons using a search test. On each trial of the search test, the monkeys were required to passively maintain their gaze on a red fixation target square (size = 0.17°) presented in the center of the monitor and visible during the entire trial. A trial started with the onset of the fixation target. After 300 ms of stable fixation, a stimulus was presented for 300 ms. To complete a trial and obtain a fluid reward, the monkeys had to maintain fixation for 300 ms poststimulus. The different images were presented foveally in a random manner. Once the spiking activity of a responsive neuron was well isolated or at a responsive multi-unit site, the search test was used to select two stimuli, with one of the two stimuli evoking a strong response (effective stimulus, labeled A) and the other little or no response (ineffective stimulus, labeled B). The two stimuli were selected on-line from the pool of 52 color images by inspection of post-stimulus time histograms (PSTH) of the responses of the neuron or multi-unit site to each of the stimuli.

In the majority of recordings, we added 6 conditions in which during the period that the adapter stimulus would have been presented, only the fixation target was shown. Thus, these trials lasted for the same durations as those in which an adapter was presented but the test stimulus was preceded only by presentation of the fixation target (see Fig. 1). Thus, these 6 additional conditions consisted of presentations of A or B as test stimulus following one of the 3 durations (300, 1500 and 3000 ms) without presentation of the adapter. The timing was exactly the same as for the other corresponding 12 conditions. The 18 conditions (12 + 6) of the full adaptation test were presented in random order in blocks of 36 unaborted trials, with a minimum of 6 unaborted trials per condition.

In a separate set of recordings, we manipulated the stimulus position independently for the adapter and test stimuli: the adapter and test stimuli were presented at the same or a different, non-overlapping position. Each of the two stimuli could be shown at the 5° eccentricity in the lower or upper visual field. We employed two adapter durations: 300 and 3000 ms. Thus, this test included 32 conditions (2 positions × 2 stimulus types (adapter versus test) × 2 stimuli (A and B)) × 2 trial types × 2 adapter durations). The 32 conditions were presented in random order in blocks of 64 unaborted trials, with a minimum of 6 trials/condition (median = 10 trials/condition). We tested only single units in these recording sessions.

### Data analysis

For each recorded neuron and condition, we computed the mean firing rate to the adapter and test stimuli. Only unaborted trials were analyzed. Responses to the stimuli were computed within a 300 ms long analysis window starting at 60 ms after stimulus onset. In all experiments, the onset of the adapter and test stimuli were detected by a photodiode and those timings were employed to align the neural activity to the stimuli for the two stimulus types separately. For each neuron and condition we computed an adaptation contrast index:

$( n e t r e s p o n s e t o a d a p t e r − n e t r e s p o n s e t o t e s t ) / ( | n e t r e s p o n s e t o a d a p t e r | + | n e t r e s p o n s e t o t e s t | ) .$

The net responses were computed by subtracting the baseline firing rate from the firing rate in the 300 ms analysis window. The baseline activity was defined as the mean firing rate in a 100 ms long interval that ended at the onset of the adapter stimulus.

PSTHs were computed for each neuron and condition by averaging the net firing rate in bins of 20 ms, aligned on stimulus onset. Then, population PSTHs were created by averaging the normalized net firing rate across neurons per condition. Normalization was performed per neuron by dividing the net firing rate by the maximum firing rate in a 20 ms bin across the AA repetition and the BA alternation trials. The BB and AB trials were not further analyzed: these trials served only to make the test stimulus identity unpredictable to the monkey. We computed the significance of the difference between the population PSTHs of different conditions by a Wilcoxon signed rank test applied to the net firing rate in each of 20 ms bins from 60 ms to 300 ms after stimulus onset. The p values were corrected for multiple comparisons (bins) using the Benjamini and Hochberg13 False Discovery Rate method (q < 0.05 was considered as statistically significant). The standard errors of the mean responses in the population PSTHs were computed following the procedure by Loftus and Masson14 which removes the variance due to the differences in the overall mean response across neurons or sites.

LFPs were filtered offline with a digital 50-Hz notch filter (48–52 Hz fourth-order Butterworth FIR filter; Fieldtrip Toolbox, F.C. Donders Centre for Cognitive Neuroimaging, Nijmegen, The Netherlands) to remove 50 Hz. Trials in which the signal was <1% or >99% of the total input range were excluded. We employed the same method for spectral analysis of the LFP as De Baene and Vogels6 and Kaliukhovich and Vogels10. By convolving single-trial data using complex Morlet wavelets15 and taking the square of the convolution between the wavelet and signal, the time-varying power of the signal for every frequency was obtained. Averaging spectral maps (power as a function of frequency and time) across trials for a given condition and site produced a spectral map of that condition and site. The complex Morlet wavelets had a constant center frequency-spectral bandwidth ratio $f 0 / σ f$ of 7, with $f 0$ ranging from 15 to 170 Hz in steps of 1 Hz. The spectral maps of the sites were normalized at each frequency by the average power within the baseline window of 300 ms before adapter onset.

Since repetition suppression in IT LFPs is consistently present for only the high frequency power6, 9, we quantified the adaptation effects by pooling the normalized power for the gamma frequencies between 70 and 170 Hz. Adaptation contrast indices were computed using the thus averaged gamma power within a temporal window that ranged between 50 and 350 ms post stimulus onset.

## Results

Closer examination of the responses for the different adapter durations in Figs 2, 3 and 4, which represent independent neuronal samples, shows a consistently different time course of the response to the test stimuli in repetition trials amongst the three adapter durations, despite similar average responses. This can be seen clearly when pooling all the available single unit data for the 300 and 3000 ms adapter durations of the (same position) repetition and alternation trials (Fig. 5A). Although the peak responses to the test stimuli following the different adapter durations in repetition trials were indistinguishable, there was a significantly stronger initial test response for the 300 compared to the 3000 ms adapter duration followed by a significantly more sustained, stronger response for the longer adapter duration condition. The significances shown in Fig. 5A were based on false discovery rate corrected Wilcoxon signed rank tests using a bin width of 20 ms. There was no significant late enhancement of the response to the test stimuli in alternation trials for the 3000 compared to the 300 ms adapter durations and also no difference between the response onset time course for the two duration conditions in the alternation trials. Thus, the difference between the time courses of the test responses for the two adapter duration conditions were not related to the timing of the stimuli per se, but were specific to stimulus repetition. Note that in this larger sample of neurons (n = 162), median adaptation indices, computed using a 300 ms analysis window, were again similar and statistically indistinguishable for the 300 (median = 0.20) and 3000 ms long adapter durations (median = 0.21; Wilcoxon signed rank test: p = 0.11).

These data show that adaptation duration has a peculiar effect on the time course of the response to the repeated stimulus without affecting the overall spike count in a window that encompasses the test duration. The absence of an overall effect of adapter duration on overall spike count can be seen also when computing the median number of spikes during the test stimulus presentation for all 92 single neurons that were tested with 3 adapter durations (Fig. 6, bottom). These highly similar spike counts during the test stimulus presentations are in strong contrast with the increasingly higher spike counts, integrated during the adapter stimulus presentation, with increasing adapter duration (Fig. 6, top). Thus, the fivefold increase in spike counts obtained by prolonging the adapter presentation had no effect on the overall spike count computed during the test presentation (compare top and bottom panels of Fig. 6).

Together with spiking activity, we measured also local field potentials. Time frequency analysis using Morlet wavelets (see Methods) of the LFPs showed evidence of repetition suppression for frequencies above 70 Hz in both animals. Figure 7 shows the average normalized power in the 70–170 Hz frequency band for the adapter and test stimuli in repetition and alternation trials and this for the 3 adapter duration conditions separately. The average high-frequency band power, measured in a window of 60 to 300 ms post-stimulus onset, was smaller during the test presentations in repetition trials, compared with the adapter, and this held true for each adapter duration (Wilcoxon sign rank tests; all P’s < 10−9; N = 92 sites). As for the spiking activity, the power to the repeated test stimulus increased faster after the 300 ms compared with the longer adapter durations. Thus, the average power to the repeated test stimulus, measured in a window from 60 to 100 ms post-stimulus onset, was significantly greater for the 300 ms compared to the 3000 ms adapter duration (Wilcoxon signed rank test; P < 0.0005). However, there was greater power with increasing adapter durations in the later window that ranged from 140 to 200 ms post-stimulus onset (difference 300 versus 3000 adapter duration: Wilcoxon signed rank test; P < 10−6). Note that these early and late analysis windows correspond to the time at which the spiking activity showed significant differences between the 300 and 3000 ms (Fig. 5A).

## Discussion

We expected that increasing the duration of an adapter stimulus would increase the degree of repetition suppression of the spiking activity in macaque IT. However, we found that large variations in adapter duration (in the range of 300–3000 ms) did not affect the overall degree of repetition suppression. This held true when adapter and test stimuli were presented at identical or different visual field positions. The invariance of overall repetition suppression with adapter duration did not result from a ceiling effect, since the repeated presentation of a short adapter stimulus produced stronger repetition suppression. An analysis of the time course of the spiking activity and the LFP gamma power showed changes in the response to the repeated stimulus with adapter duration: the initial part of the response to the repeated stimulus was greater for shorter compared to longer adapter durations, while the opposite was true for the later part of the response. These changes in time course of the test response were consistent across different independent experiments. Furthermore, we found that IT neurons that responded in a transient manner to a long duration adapter stimulus also showed stronger repetition suppression.

Previous studies of awake macaque IT repetition suppression varied adapter durations either within a smaller range and across neurons3 or for shorter durations5. Sawamura and coworker3 tested partially overlapping sets of neurons for durations varying between 300 and 900 ms and observed similar repetition suppression for these durations. Although this finding agrees with that of the present study, the absence of an effect of adapter duration in that study needs to be interpreted with caution since comparisons were made across small only partially overlapping samples of neurons. Also, the adapter duration range was quite low. Liu et al.5 found also similar time courses and degree of repetition suppression for a 150 and 500 ms adapter duration, tested in different neuronal samples. However, the degree of repetition suppression was significantly smaller for a short 80 ms adapter compared to the 500 ms duration. Combined with the result of the present study, this suggests that very short adapter durations produce less repetition suppression than longer ones, but this overall increase in repetition suppression with adapter duration appears to be limited to durations shorter than 150 ms (according to)5 or 300 ms (present study).

Zago and coworkers18 varied systematically exposure durations, ranging between 40 and 1900 ms, in a fMRI repetition priming paradigm. They found that repetition suppression in occipito-temporal cortex peaked with a prime exposure duration of 250 ms. They found no difference between 350 and 1900 exposure durations. Although these findings agree with our study (and that of Liu et al.5 observation of less repetition suppression for a 80 ms adapter), the studies are difficult to compare because of the highly different paradigms: we employed a short 300 ms ISI with no intervening stimuli while in Zago et al.’s18 priming paradigm intervening stimuli (including a mask following the stimulus) between repetitions were present18, leading to variable and long intervals between the first and second presentation of the same stimulus (>2 s up to minutes).

A recent human ERP study examined the effect of adapter duration, ranging between 200 to 5000 ms, on the response evoked by the repeated presentation of faces19. Their design, using static adapters and a short 500 ms ISI, was comparable to ours. Since for most of our neurons, the two adapter stimuli belonged to different categories, the scrambled face versus face adapter comparison (their “generic” adaptation)19 comes the closest to our stimulus conditions. Zimmer and coworkers19 reported that the latter stimulus specific adaptation increased with adapter duration for the early P100 and the late P2 ERP components but less clearly for the N170 and the late N250 components. The current source(s) in the brain of these different ERP components are uncertain which makes a comparison with our IT data difficult. Nonetheless, it is tempting to speculate that the duration effect of the P100 component is related to the stronger adaptation effect with increasing adapter duration at 60–100 ms after stimulus onset in monkey IT. However, the later reversal of the adapter duration effect that we observed does not appear to be present in these human ERP data19.

The above qualitative framework assumes that repetition suppression results from fatigue mechanisms like response-driven fatigue and/or synaptic or input fatigue6, 11. However, it has been proposed that repetition suppression reflects top-down expectations about stimulus repetition26. Although evidence for this proposal is lacking at the single unit level in IT10, it could explain why the overall level of repetition suppression is similar for different adapter durations since expectations about repetition are unlikely to depend on stimulus duration. Note that to avoid that our adaptation protocol induced expectations of repetition we had an equal number of repetition and alternation trials. Thus, when repetition suppression reflected a fulfilled expectation of repetition, as postulated by Summerfield and coworkers26 then this expectation must be a rather fixed “prior” that is not malleable to changes in repetition probability (which was 0.50 in our paradigm). Furthermore, as far as we know, such a predictive coding account of repetition suppression does not predict the differential effects of adapter duration on repetition suppression during the course of the response that we observed in the present study.

To summarize, we found that increasing the duration of a static adapter beyond 300 ms and up to 3000 ms does not produce an overall increase of the degree of repetition suppression, despite the increase in the number of spikes during the adaptation period. However, prolonging adapter duration resulted in an increased repetition suppression during the initial phase of the response but a decreased repetition suppression during the later phase of the response to the test stimulus. The degree of repetition suppression for the test stimulus correlated with the relative degree of reduction of the sustained phase of the response, suggesting at least overlapping mechanisms that drive the response during the adapter duration and the subsequent repetition suppression to a delayed test stimulus.

### Data availability

The data and analysis code that support the findings of this study are available from the authors upon reasonable request.

Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

## References

1. 1.

Miller, E. K., Li, L. & Desimone, R. A neural mechanism for working and recognition memory in inferior temporal cortex. Science 254, 1377 (1991).

2. 2.

Vogels, R., Sary, G. & Orban, G. A. How task-related are the responses of inferior temporal neurons? Vis. Neurosci 12, 207–14 (1995).

3. 3.

Sawamura, H., Orban, G. A. & Vogels, R. Selectivity of Neuronal Adaptation Does Not Match Response Selectivity: A Single-Cell Study of the fMRI Adaptation Paradigm. Neuron 49, 307–18 (2006).

4. 4.

McMahon, D. B. T. & Olson, C. R. Repetition Suppression in Monkey Inferotemporal Cortex: Relation to Behavioral Priming. J. Neurophysiol. 97, 3532 (2007).

5. 5.

Liu, Y., Murray, S. O. & Jagadeesh, B. Time Course and Stimulus Dependence of Repetition-Induced Response Suppression in Inferotemporal Cortex. J. Neurophysiol. 101, 418 (2009).

6. 6.

De Baene, W. & Vogels, R. Effects of Adaptation on the Stimulus Selectivity of Macaque Inferior Temporal Spiking Activity and Local Field Potentials. Cerb. Cortex. 20, 2145–65 (2010).

7. 7.

Kaliukhovich, D. A. & Vogels, R. Divisive Normalization Predicts Adaptation-Induced Response Changes in Macaque Inferior Temporal Cortex. J. Neurosci. 36, 6116 (2016).

8. 8.

Kaliukhovich, D. A. & Vogels, R. Neurons in Macaque Inferior Temporal Cortex Show No Surprise Response to Deviants in Visual Oddball Sequences. J. Neurosci. 34, 12801–15 (2014).

9. 9.

Kaliukhovich, D. A. & Vogels, R. Stimulus repetition affects both strength and synchrony of macaque inferior temporal cortical activity. J. Neurophysiol. 107, 3509–27 (2012).

10. 10.

Kaliukhovich, D. A. & Vogels, R. Stimulus Repetition Probability Does Not Affect Repetition Suppression in Macaque Inferior Temporal Cortex. Cereb. Cortex. 21, 1547–58 (2011).

11. 11.

Vogels, R. Sources of adaptation of inferior temporal cortical responses. Cortex 80, 185–95 (2016).

12. 12.

Kovacs, G. & Vogels, R. When does repetition suppression depend on repetition probability? Front. Hum. Neurosci 8, 685 (2014).

13. 13.

Benjamini, Y. & Hochberg, Y. Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. Journal of the Royal Statistical Society. Series B (Methodological) 57, 289–300 (1995).

14. 14.

Loftus, G. R. & Masson, M. E. J. Using confidence intervals in within-subject designs. Psychonomic Bulletin & Review 1, 476–90 (1994).

15. 15.

Tallon-Baudry, C., Bertrand, O., Delpuech, C. & Pernier, J. Oscillatory gamma-band (30–70 Hz) activity induced by a visual search task in humans. J. Neurosci. 17, 722–34 (1997).

16. 16.

Patterson, C. A., Wissig, S. C. & Kohn, A. Distinct Effects of Brief and Prolonged Adaptation on Orientation Tuning in Primary Visual Cortex. J. Neurosci. 33, 532–43 (2013).

17. 17.

Fang, F., Murray, S. O. & He, S. Duration-Dependent fMRI Adaptation and Distributed Viewer-Centered Face Representation in Human Visual Cortex. Cereb. Cortex. 17, 1402–11 (2007).

18. 18.

Zago, L., Fenske, M. J., Aminoff, E. & Bar, M. The Rise and Fall of Priming: How Visual Exposure Shapes Cortical Representations of Objects. Cereb. Cortex 15, 1655–65 (2005).

19. 19.

Zimmer, M., Zbant, A., Nemeth, Kl & Kovacs, G. Adaptation Duration Dissociates Category-, Image-, and Person-Specific Processes on Face-Evoked Event-Related Potentials. Front. Psychol. 6, 1945 (2015).

20. 20.

Abolafia, J. M., Vergara, R., Arnold, M. M., Reig, R. & Sanchez-Vives, M. V. Cortical Auditory Adaptation in the Awake Rat and the Role of Potassium Currents. Cereb.Cortex. 21, 977–90 (2011).

21. 21.

Zucker, R. S. & Regehr, W. G. Short-term synaptic plasticity. Annu. Rev. Physiol 64, 355–405 (2002).

22. 22.

Fioravante, D. & Regehr, W. G. Short-term forms of presynaptic plasticity. Curr. Opin. Neurobiol. 21, 269–74 (2011).

23. 23.

Benda, J. & Herz, A. V. M. A Universal Model for Spike-Frequency Adaptation. Neural Comput. 15, 2523–64 (2003).

24. 24.

Lisberger, S. G. & Movshon, J. A. Visual motion analysis for pursuit eye movements in area MT of macaque monkeys. J. Neurosci. 19, 2224–2246 (1999).

25. 25.

Priebe, N. J., Churchland, M. M. & Lisberger, S. G. Constraints on the source of short-term motion adaptation in macaque area MT. I. The role of input and intrinsic mechanisms. J. Neurophysiol. 88, 354–369 (2002).

26. 26.

Summerfield, C., Trittschuh, E. H., Monti, J. M., Mesulam, M. M. & Egner, T. Neural repetition suppression reflects fulfilled perceptual expectations. Nat. Neurosci. 11, 1004–06 (2008).

## Acknowledgements

This work was supported by Fonds voor Wetenschappelijk Onderzoek Vlaanderen, Interuniversitaire Attractiepool and Programma Financiering (PF 10/008) and the European Community’s Seventh Framework Programme FP7/2007-2013 under grant agreement number PITN-GA-2008-290011 (ABC). We thank P. Kayenbergh, G. Meulemans, I. Puttemans, Christophe Ulens, M. De Paep, W. Depuydt, and S. Verstraeten for technical assistance.

## Author information

### Affiliations

1. #### Laboratorium voor Neuro- en Psychofysiologie, Department of Neurosciences, KULeuven, Leuven, Belgium

•  & Rufin Vogels

### Contributions

P.K. and R.V. designed the study. P.K. performed the experiments, analyzed the data and prepared all figures. R.V. and P.K. wrote the main manuscript text.

### Competing Interests

The authors declare that they have no competing interests.

### Corresponding author

Correspondence to Rufin Vogels.