A neural mechanism for time-window separation resolves ambiguity of adaptive coding

Neural adaptation

The term 'neural adaptation' refers to the common phenomenon of decaying neuronal activities in response to repeated or prolonged stimulation. Many different roles of adaptation in neural computations have been discussed. On a single-cell level adaptation introduces a high-pass filter operation as a basic element for predictive coding. Interactions of adaptation processes with nonlinearities are key to many more computations including generation of invariances, stimulus selectivity, denoising, and sparsening. Neural adaptation is observed all the way along neuronal pathways from the sensory periphery to the motor output and adaptation usually gets stronger at higher levels. Non-adapting neurons or neurons that increase their sensitivity are rare exceptions. What computations arise by repeated adaptation mechanisms along a processing pathway? After giving some background on neural adaptation, underlying mechanisms, dynamics, and resulting filter properties, I will discuss computational properties of four examples of serial and parallel adaptation processes, demonstrating that adaptation acts together with other mechanisms, in particular threshold nonlinearities, to eventually compute meaningful perceptions. Python code and further details of the simulations illustrating this primer are available at https://github.com/janscience/adaptationprimer.

Sensitivity to interaural time differences and localization accuracy in cochlear implant users with combined electric-acoustic stimulation

OBJECTIVES

In this study, localization accuracy and sensitivity to acoustic interaural time differences (ITDs) in subjects using cochlear implants with combined electric-acoustic stimulation (EAS) were assessed and compared with the results of a normal hearing control group.

METHODS

Eight CI users with EAS (2 bilaterally implanted, 6 unilaterally implanted) and symmetric binaural acoustic hearing and 24 normal hearing subjects participated in the study. The first experiment determined mean localization error (MLE) for different angles of sound incidence between ± 60° (frontal and dorsal presentation). The stimuli were either low-pass, high-pass or broadband noise bursts. In a second experiment, just noticeable differences (JND) of ITDs were measured for pure tones of 125 Hz, 250 Hz and 500 Hz (headphone presentation).

RESULTS

Experiment 1: MLE of EAS subjects was 8.5°, 14.3° and 14.7°, (low-, high-pass and broadband stimuli respectively). In the control group, MLE was 1.8° (broadband stimuli). In the differentiation between sound incidence from front and back, EAS subjects performed on chance level. Experiment 2: The JND-ITDs were 88.7 μs for 125 Hz, 48.8 μs for 250 Hz and 52.9 μs for 500 Hz (EAS subjects). Compared to the control group, JND-ITD for 125 Hz was on the same level of performance. No statistically significant correlation was found between MLE and JND-ITD in the EAS cohort.

CONCLUSIONS

Near to normal ITD sensitivity in the lower frequency acoustic hearing was demonstrated in a cohort of EAS users. However, in an acoustic localization task, the majority of the subjects did not reached the level of accuracy of normal hearing. Presumably, signal processing time delay differences between devices used on both sides are deteriorating the transfer of precise binaural timing cues.

Circuit and Cellular Mechanisms Facilitate the Transformation from Dense to Sparse Coding in the Insect Olfactory System

Transformations between sensory representations are shaped by neural mechanisms at the cellular and the circuit level. In the insect olfactory system, the encoding of odor information undergoes a transition from a dense spatiotemporal population code in the antennal lobe to a sparse code in the mushroom body. However, the exact mechanisms shaping odor representations and their role in sensory processing are incompletely identified. Here, we investigate the transformation from dense to sparse odor representations in a spiking model of the insect olfactory system, focusing on two ubiquitous neural mechanisms: spike frequency adaptation at the cellular level and lateral inhibition at the circuit level. We find that cellular adaptation is essential for sparse representations in time (temporal sparseness), while lateral inhibition regulates sparseness in the neuronal space (population sparseness). The interplay of both mechanisms shapes spatiotemporal odor representations, which are optimized for the discrimination of odors during stimulus onset and offset. Response pattern correlation across different stimuli showed a nonmonotonic dependence on the strength of lateral inhibition with an optimum at intermediate levels, which is explained by two counteracting mechanisms. In addition, we find that odor identity is stored on a prolonged timescale in the adaptation levels but not in the spiking activity of the principal cells of the mushroom body, providing a testable hypothesis for the location of the so-called odor trace.

Temporal integration of conflicting directional cues in sound localization

Sound localization is fundamental to hearing. In nature, sound degradation and noise erode directional cues and can generate conflicting directional perceptions across different subcomponents of sounds. Little is known about how sound localization is achieved in the face of conflicting directional cues in non-human animals, although this is relevant for many species in which sound localization in noisy conditions mediates mate finding or predator avoidance. We studied the effects of conflicting directional cues in male grasshoppers, Chorthippus biguttulus, which orient towards signaling females. We presented playbacks varying in the number and temporal position of song syllables providing directional cues in the form of either time or amplitude differences between two speakers. Males oriented towards the speaker broadcasting a greater number of leading or louder syllables. For a given number of syllables providing directional information, syllables with timing differences at the song's beginning were weighted most heavily, while syllables with intensity differences were weighted most heavily when they were in the middle of the song. When timing and intensity cues conflicted, the magnitude and temporal position of each cue determined their relative influence on lateralization, and males sometimes quickly corrected their directional responses. We discuss our findings with respect to similar results from humans.

Multivariate female preference tests reveal latent perceptual biases

The question of why males of many species produce elaborate mating displays has now been largely resolved: females prefer to mate with males that produce such displays. However, the question of why females prefer such displays has been controversial, with an emerging consensus that such displays often provide information to females about the direct fitness benefits that males provide to females and/or the indirect fitness benefits provided to offspring. Alternative explanations, such as production of arbitrarily attractive sons or innate pre-existing female sensory or perceptual bias, have also received support in certain taxa. Here, we describe multivariate female preference functions for male acoustic traits in two chirping species of field crickets with slow pulse rates; our data reveal cryptic female preferences for long trills that have not previously been observed in other chirping species. The trill preferences are evolutionarily pre-existing in the sense that males have not (yet?) exploited them, and they coexist with chirp preferences as alternative stable states within female song preference space. We discuss escape from neuronal adaptation as a possible mechanism underlying such latent preferences.

Optimal coding and neuronal adaptation in economic decisions

During economic decisions, offer value cells in orbitofrontal cortex (OFC) encode the values of offered goods. Furthermore, their tuning functions adapt to the range of values available in any given context. A fundamental and open question is whether range adaptation is behaviorally advantageous. Here we present a theory of optimal coding for economic decisions. We propose that the representation of offer values is optimal if it ensures maximal expected payoff. In this framework, we examine offer value cells in non-human primates. We show that their responses are quasi-linear even when optimal tuning functions are highly non-linear. Most importantly, we demonstrate that for linear tuning functions range adaptation maximizes the expected payoff. Thus value coding in OFC is functionally rigid (linear tuning) but parametrically plastic (range adaptation with optimal gain). Importantly, the benefit of range adaptation outweighs the cost of functional rigidity. While generally suboptimal, linear tuning may facilitate transitive choices.

Complex object motion represented by context-dependent correlated activity of visual interneurones

Accurate and adaptive encoding of complex, dynamic visual information is critical for the survival of many animals. Studies across a range of taxa have investigated behavioral and neuronal responses to objects that represent a threat, such as a looming object approaching along a direct collision course. By investigating neural mechanisms of avoidance behaviors through recording multineuronal activity, it is possible to better understand how complex visual information is represented in circuits that ultimately drive behaviors. We used multichannel electrodes to record from the well-studied locust nervous system to explore how object motion is reflected in activity of correlated neural activity. We presented locusts (Locusta migratoria) with objects that moved along one of 11 unique trajectories and recorded from descending interneurons within the ventral nerve cord. Spike sorting resulted in 405 discriminated units across 20 locusts and we found that 75% of the units responded to some form of object motion. Dimensionality reduction through principal component (PCA) and dynamic factor (DFA) analyses revealed population vector responses within individuals and common firing trends across the pool of discriminated units, respectively. Population vector composition (PCA) varied with the stimulus and common trends (DFA) showed unique tuning related to changes in the visual size and trajectory of the object through time. These findings demonstrate that this well-described collision detection system is more complex than previously envisioned and will drive future experiments to explore fundamental principles of how visual information is processed through context-dependent dynamic ensembles of neurons to initiate and control complex behavior.

Exploring the hidden landscape of female preferences for complex signals

A major challenge in evolutionary biology is explaining the origins of complex phenotypic diversity. In animal communication, complex signals may evolve from simpler signals because novel signal elements exploit preexisting biases in receivers' sensory systems. Investigating the shape of female preference functions for novel signal characteristics is a powerful, but underutilized, method to describe the adaptive landscape potentially guiding complex signal evolution. We measured female preference functions for characteristics of acoustic appendages added to male calling songs in the grasshopper Chorthippus biguttulus, which naturally produces only simple songs. We discovered both hidden preferences for and biases against novel complex songs, and identified rules governing song attractiveness based on multiple characteristics of both the base song and appendage. The appendage's temporal position and duration were especially important: long appendages preceding the song often made songs less attractive, while following appendages were neutral or weakly attractive. Appendages had stronger effects on songs of shorter duration, but did not restore the attractiveness of very unattractive songs. We conclude that sensory biases favor, within predictable limits, the evolution of complex songs in grasshoppers. The function-valued approach is an important tool in determining the generality of these limits in other taxa and signaling modalities.

Adaptation to second order stimulus features by electrosensory neurons causes ambiguity

Understanding the coding strategies used to process sensory input remains a central problem in neuroscience. Growing evidence suggests that sensory systems process natural stimuli efficiently by ensuring a close match between neural tuning and stimulus statistics through adaptation. However, adaptation causes ambiguity as the same response can be elicited by different stimuli. The mechanisms by which the brain resolves ambiguity remain poorly understood. Here we investigated adaptation in electrosensory pyramidal neurons within different parallel maps in the weakly electric fish Apteronotus leptorhynchus. In response to step increases in stimulus variance, we found that pyramidal neurons within the lateral segment (LS) displayed strong scale invariant adaptation whereas those within the centromedial segment (CMS) instead displayed weaker degrees of scale invariant adaptation. Signal detection analysis revealed that strong adaptation in LS neurons significantly reduced stimulus discriminability. In contrast, weaker adaptation displayed by CMS neurons led to significantly lesser impairment of discriminability. Thus, while LS neurons display adaptation that is matched to natural scene statistics, thereby optimizing information transmission, CMS neurons instead display weaker adaptation and would instead provide information about the context in which these statistics occur. We propose that such a scheme is necessary for decoding by higher brain structures.

Adaptive Spike Threshold Enables Robust and Temporally Precise Neuronal Encoding

Neural processing rests on the intracellular transformation of information as synaptic inputs are translated into action potentials. This transformation is governed by the spike threshold, which depends on the history of the membrane potential on many temporal scales. While the adaptation of the threshold after spiking activity has been addressed before both theoretically and experimentally, it has only recently been demonstrated that the subthreshold membrane state also influences the effective spike threshold. The consequences for neural computation are not well understood yet. We address this question here using neural simulations and whole cell intracellular recordings in combination with information theoretic analysis. We show that an adaptive spike threshold leads to better stimulus discrimination for tight input correlations than would be achieved otherwise, independent from whether the stimulus is encoded in the rate or pattern of action potentials. The time scales of input selectivity are jointly governed by membrane and threshold dynamics. Encoding information using adaptive thresholds further ensures robust information transmission across cortical states i.e. decoding from different states is less state dependent in the adaptive threshold case, if the decoding is performed in reference to the timing of the population response. Results from in vitro neural recordings were consistent with simulations from adaptive threshold neurons. In summary, the adaptive spike threshold reduces information loss during intracellular information transfer, improves stimulus discriminability and ensures robust decoding across membrane states in a regime of highly correlated inputs, similar to those seen in sensory nuclei during the encoding of sensory information.

A neuron is a tiny computer that transforms electrical inputs into electrical outputs. While neurons have been investigated and modeled for many decades, some aspects remain elusive. Recently, it was demonstrated that the membrane (voltage) state of a neuron determines its threshold to spiking. In the present study we asked, what are the consequences of this dependence for the computation the neuron performs. We find that this so called adaptive threshold allows neurons to be more focused on inputs which arrive close in time with other inputs. Also, it allows neurons to represent their information more robustly, such that a readout of their activity is less influenced by the state the brain is in. The present use of information theory provides a solid foundation for these results. We obtained the results primarily in detailed simulations, but performed neural recordings to verify these properties in real neurons. In summary, an adaptive spiking threshold allows neurons to specifically compute robustly with a focus on tight temporal correlations in their input.

Sound localization in a changing world

In natural environments, neural systems must be continuously updated to reflect changes in sensory inputs and behavioral goals. Recent studies of sound localization have shown that adaptation and learning involve multiple mechanisms that operate at different timescales and stages of processing, with other sensory and motor-related inputs playing a key role. We are only just beginning to understand, however, how these processes interact with one another to produce adaptive changes at the level of neuronal populations and behavior. Because there is no explicit map of auditory space in the cortex, studies of sound localization may also provide much broader insight into the plasticity of complex neural representations that are not topographically organized.