Sequence Of Stimuli Research Articles

Network structure influences the strength of learned neural representations

From sequences of discrete events, humans build mental models of their world. Referred to as graph learning, the process produces a model encoding the graph of event-to-event transition probabilities. Recent evidence suggests that some networks are easier to learn than others, but the neural underpinnings of this effect remain unknown. Here we use fMRI to show that even over short timescales the network structure of a temporal sequence of stimuli determines the fidelity of event representations as well as the dimensionality of the space in which those representations are encoded: when the graph was modular as opposed to lattice-like, BOLD representations in visual areas better predicted trial identity and displayed higher intrinsic dimensionality. Broadly, our study shows that network context influences the strength of learned neural representations, motivating future work in the design, optimization, and adaptation of network contexts for distinct types of learning.

Extracting the fingerprints of sequences of random rhythmic auditory stimuli from electrophysiological data.

It has been classically conjectured that the brain assigns probabilistic models to sequences of stimuli. An important issue associated with this conjecture is the identification of the classes of models used by the brain to perform this task. We address this issue by using a new clustering procedure for sets of electroencephalographic (EEG) data recorded from participants exposed to a sequence of auditory stimuli generated by a stochastic chain. This clustering procedure indicates that the brain uses the recurrent occurrences of a regular auditory stimulus in order to build a model.

Strength of Low-Frequency EEG Phase Entrainment to External Stimuli Is Associated with Fluctuations in the Brain's Internal State.

The brain attends to environmental rhythms by aligning the phase of internal oscillations. However, the factors underlying fluctuations in the strength of this phase entrainment remain largely unknown. In the present study, we examined whether the strength of low-frequency electroencephalography (EEG) phase entrainment to rhythmic stimulus sequences varied with the pupil size and posterior alpha-band power, thought to reflect the arousal level and excitability of posterior cortical brain areas, respectively. We recorded the pupil size and scalp EEG while participants carried out an intermodal selective attention task, in which they were instructed to attend to a rhythmic sequence of visual or auditory stimuli and ignore the other perceptual modality. As expected, intertrial phase coherence (ITC), a measure of entrainment strength, was larger for the task-relevant than for the task-irrelevant modality. Across the experiment, the pupil size and posterior alpha power were strongly linked with each other. Interestingly, ITC tracked both variables: larger pupil size was associated with a selective increase in entrainment to the task-relevant stimulus sequence, whereas larger posterior alpha power was associated with a decrease in phase entrainment to both the task-relevant and task-irrelevant stimulus sequences. Exploratory analyses showed that a temporal relation between ITC and posterior alpha power emerged in the time periods around pupil maxima and pupil minima. These results indicate that endogenous sources contribute distinctly to the fluctuations of EEG phase entrainment.

Computation With Sequences of Assemblies in a Model of the Brain.

Even as machine learning exceeds human-level performance on many applications, the generality, robustness, and rapidity of the brain's learning capabilities remain unmatched. How cognition arises from neural activity is the central open question in neuroscience, inextricable from the study of intelligence itself. A simple formal model of neural activity was proposed in Papadimitriou etal. (2020) and has been subsequently shown, through both mathematical proofs and simulations, to be capable of implementing certain simple cognitive operations via the creation and manipulation of assemblies of neurons. However, many intelligent behaviors rely on the ability to recognize, store, and manipulate temporal sequences of stimuli (planning, language, navigation, to list a few). Here we show that in the same model, sequential precedence can be captured naturally through synaptic weights and plasticity, and, as a result, a range of computations on sequences of assemblies can be carried out. In particular, repeated presentation of a sequence of stimuli leads to the memorization of the sequence through corresponding neural assemblies: upon future presentation of any stimulus in the sequence, the corresponding assembly and its subsequent ones will be activated, one after the other, until the end of the sequence. If the stimulus sequence is presented to two brain areas simultaneously, a scaffolded representation is created, resulting in more efficient memorization and recall, in agreement with cognitive experiments. Finally, we show that any finite state machine can be learned in a similar way, through the presentation of appropriate patterns of sequences. Through an extension of this mechanism, the model can be shown to be capable of universal computation. Taken together, these results provide a concrete hypothesis for the basis of the brain's remarkable abilities to compute and learn, with sequences playing a vital role.

Optically Modulated Nanofluidic Ionic Transistor for Neuromorphic Functions

AbstractNeuromorphic systems that can emulate the behavior of neurons have garnered increasing interest across interdisciplinary fields due to their potential applications in neuromorphic computing, artificial intelligence and brain‐machine interfaces. However, the optical modulation of nanofluidic ion transport for neuromorphic functions has been scarcely reported. Herein, inspired by biological systems that rely on ions as signal carriers for information perception and processing, we present a nanofluidic transistor based on a metal–organic framework membrane (MOFM) with optically modulated ion transport properties, which can mimic the functions of biological synapses. Through the dynamic modulation of synaptic weight, we successfully replicate intricate learning‐experience behaviors and Pavlovian associate learning processes by employing sequential optical stimuli. Additionally, we demonstrate the application of the International Morse Code with the nanofluidic device using patterned optical pulse signals, showing its encoding and decoding capabilities in information processing process. This study would largely advance the development of nanofluidic neuromorphic devices for biomimetic iontronics integrated with sensing, memory and computing functions.

A computational approach to the N-back task

The N-back task is one of the most popular paradigms for studying the cognitive mechanisms of working memory (WM). The task requires the observer to view a sequence of stimuli and judge whether the current stimulus (probe) matches the one presented N stimuli ago (target). A key phenomenon is that the intervening stimuli (distractors) interfere with task performance. Unfortunately, the classic N-back task uses complex categorical stimuli, making it unfit to quantify the effect of feature similarity on interference strength. Here, we introduce the “analog N-back task”, which utilizes stimuli varying continuously in orientation or color. This task variant enables us to measure interference strength on a continuum, providing data suitable for identifying the sources of interference using computational models. In the analog 2-back task, we found that interference increased with feature similarity between the probe and both task-relevant (1-back) and task-irrelevant (3-back) distractors. We next developed and evaluated three main models that each incorporated a Bayesian decision step and differed from an optimal non-interference model in one component only: an early-pooling model, a late-pooling model, and a substitution model. Model comparison suggests that interference emerges late in processing, most likely due to confusion between stimuli during WM retrieval. Our work puts the study of interference in the N-back task on a firmer computational footing and provides a unified framework for examining the sources of interference across domains.

Trans-retinal predictive signals of visual features are precise, saccade-specific and operate over a wide range of spatial frequencies.

Saccadic eye movements successively project the saccade target on two retinal locations: a peripheral one before the saccade, and the fovea after the saccade. Typically, performance in discriminating stimulus features changes between these two projections is very poor. However, a short (∼200 ms) blanking of the target upon saccade onset drastically improves performance, demonstrating that a precise signal of the peripheral projection is retained during the saccade. Although little is known about the nature of that transsaccadic signal, previous reports conjectured that it relies on information processed by the magnocellular system. Across two experiments, we investigated the feature blanking effect for a wide range of spatial frequencies (0.5-8 cycles per degree of visual angle, dva), stimulus sizes (1-4 dva), and eccentricities (6-10 dva). In each trial, participants executed a saccade to a high-contrast grating presented either left or right of fixation. During the saccade, the grating changed orientation (clockwise or counter-clockwise) either instantaneously or after a 200-ms blank, and participants reported the change's direction. We contrasted this saccade condition with a trans-retinal fixation condition mimicking the peripheral-then-foveal sequence of the target stimulus occurring across a saccade. Remarkably, blanking improved performance reliably for each spatial frequency, stimulus size, and eccentricity, but only in the saccade condition. Performance with blanking in saccade trials systematically exceeded performance in the fixation condition. Our results demonstrate a robust feature blanking effect across saccades, suggesting that transsaccadic processes involve low-level visual features beyond those processed in the magnocellular system.NEW & NOTEWORTHY Across a saccadic eye movement, the visual system is able to keep track of the signals carrying the visual features of a saccade target. We provide evidence that these signals are sensitive to a wide range of stimulus sizes, can use a wide range spatial frequencies channels and, operate at various saccade amplitudes. Our results suggest an underlying mechanism operating beyond the magnocellular pathway that is contingent to saccade execution.

Temporal attention amplifies stimulus information in fronto-cingulate cortex at an intermediate processing stage.

The human brain faces significant constraints in its ability to process every item in a sequence of stimuli. Voluntary temporal attention can selectively prioritize a task-relevant item over its temporal competitors to alleviate these constraints. However, it remains unclear when and where in the brain selective temporal attention modulates the visual representation of a prioritized item. Here, we manipulated temporal attention to successive stimuli in a two-target temporal cueing task, while controlling for temporal expectation with fully predictable stimulus timing. We used magnetoencephalography and time-resolved decoding to track the spatiotemporal evolution of stimulus representations in human observers. We found that temporal attention enhanced the representation of the first target around 250 ms after target onset, in a contiguous region spanning left frontal cortex and cingulate cortex. The results indicate that voluntary temporal attention recruits cortical regions beyond the ventral stream at an intermediate processing stage to amplify the representation of a target stimulus. This routing of stimulus information to anterior brain regions may provide protection from interference in visual cortex by a subsequent stimulus. Thus, voluntary temporal attention may have distinctive neural mechanisms to support specific demands of the sequential processing of stimuli.

Interbrain neural correlates of self and other integration in joint statistical learning

While statistical learning is often studied individually, its collective representation through self-other integration remains unclear. This study examines dynamic self-other integration and its multi-brain mechanism using simultaneous recordings from dyads. Participants (N = 112) each repeatedly responded to half of a fixed stimulus sequence with either an active partner (joint context) or a passive observer (baseline context). Significant individual statistical learning was evident in the joint context, characterized by decreased reaction time (RT) and intra-brain neural responses, followed by a quadratic trend (i.e., first increasing and then decreasing) upon insertion of an interference sequence. More importantly, Brain-to-Brain Coupling (BtBC) in the theta band also showed learning and modulation-related trends, with its slope negatively and positively correlating with the slopes of RT and intra-brain functional connectivity, respectively. These results highlight the dynamic nature of self-other integration in joint statistical learning, with statistical regularities implicitly and spontaneously modulating this process. Notably, the BtBC serves as a key neural correlate underlying the dynamics of self-other integration.

A receptor-inactivation model for single-celled habituation in Stentor coeruleus.

The single-celled ciliate Stentor coeruleus demonstrates habituation to mechanical stimuli, showing that even single cells can manifest a basic form of learning. Although the ability of Stentor to habituate has been extensively documented, the mechanism of learning is currently not known. Here we take a bottom-up approach and investigate a simple biochemistry-based model based on prior electrophysiological measurements in Stentor along with general properties of receptor molecules. In this model, a mechanoreceptor senses the stimulus and leads to channel opening to change membrane potential, with a sufficient change in polarization triggering an action potential that drives contraction. Receptors that are activated can become internalized, after which they can either be degraded or recycled back to the cell surface. This activity-dependent internalization provides a potential means for the cell to learn. Stochastic simulations of this model confirm that it is capable of showing habituation similar to what is seen in actual Stentor cells, including the lack of dishabituation by strong stimuli and the apparently step-like response of individual cells during habituation. The model also can account for several habituation hallmarks that a previous two-state Markov model could not, namely, the dependence of habituation rate on stimulus magnitude, which had to be added onto the two state model but arises naturally in the receptor inactivation model; the rate of response recovery after cessation of stimulation; the ability of high frequency stimulus sequences to drive faster habituation that results in a lower response probability, and the potentiation of habituation by repeated rounds of training and recovery. The model makes the prediction that application of high force stimuli that do not normally habituate should drive habituation to weaker stimuli due to decrease in the receptor number, which serves as an internal hidden variable. We confirmed this prediction using two new sets of experiments involving alternation of weak and strong stimuli. Furthermore, the model predicts that training with high force stimuli delays response recovery to low force stimuli, which aligns with our new experimental data. The model also predicts subliminal accumulation, wherein continuation of training even after habituation has reached asymptotic levels should lead to delayed response recovery, which was also confirmed by new experiments. The model is unable to account for the phenomenon of rate sensitivity, in which habituation caused by higher frequency stimuli is more easily reversed leading to a frequency dependence of response recovery. Such rate sensitivity has not been reported in Stentor . Here we carried out a new set of experiments which are consistent with the model's prediction of the lack of rate sensitivity. This work demonstrates how a simple model can suggest new ways to probe single-cell learning at an experimental level. Finally, we interpret the model in terms of a kernel estimator that the cell may use to guide its decisions about how to response to new stimuli as they arise based on information, or the lack thereof, from past stimuli.

Stereodivergent Macrocyclization in Dynamic Chiral Confinement.

The absolute and relative configurations of a macrocyclic natural product bearing multiple chirality have a crucial influence on its physical and biological properties. Nevertheless, their preparation with full stereocontrol remains largely unexplored in synthetic community. Here, we show a stereodivergent macrocyclization under dynamic chiral confinement in which the stepwise chirality switching of a chiral space directs complete stereocontrol. To confine a substrate in a dynamic chiral space, we used a chiral capsule enclosing a substrate which is collectively switchable in the chirality in response to external stimuli, but their conformations are firmly fixed by subsequent self-assembly. The consecutive chirality switching enables the confined reactions of an enclosed achiral substrate to sequentially install diverse chirality on a macrocycle product with full stereocontrol in a single pot operation, thus changing only a sequence of physical stimuli enables access to a remarkable stereodivergence in a macrocyclization process.

Perturbation context in paced finger tapping tunes the error-correction mechanism

Sensorimotor synchronization (SMS) is the mainly specifically human ability to move in sync with a periodic external stimulus, as in keeping pace with music. The most common experimental paradigm to study its largely unknown underlying mechanism is the paced finger-tapping task, where a participant taps to a periodic sequence of brief stimuli. Contrary to reaction time, this task involves temporal prediction because the participant needs to trigger the motor action in advance for the tap and the stimulus to occur simultaneously, then an error-correction mechanism takes past performance as input to adjust the following prediction. In a different, simpler task, it has been shown that exposure to a distribution of individual temporal intervals creates a “temporal context” that can bias the estimation/production of a single target interval. As temporal estimation and production are also involved in SMS, we asked whether a paced finger-tapping task with period perturbations would show any time-related context effect. In this work we show that a perturbation context can indeed be generated by exposure to period perturbations during paced finger tapping, affecting the shape and size of the resynchronization curve. Response asymmetry is also affected, thus evidencing an interplay between context and intrinsic nonlinearities of the correction mechanism. We conclude that perturbation context calibrates the underlying error-correction mechanism in SMS.

Do smaller P300 amplitudes in schizophrenia result from larger variability in temporal processing?

The P3b is a prominent event-related potential (ERP) with maximal amplitude between 250 ms and 500 ms after the onset of a rare target stimulus within a sequence of standard non-target stimuli (oddball paradigm). Several studies found reduced P3b amplitudes in patients with schizophrenia compared to neurotypicals. Our work and the literature suggest that temporal imprecision may play a large pathophysiological role in schizophrenia. Here, we investigated whether reduced P3b amplitudes result from reduced neural activity (power) or temporal imprecision (inter-trial phase coherence; ITC) in delta and theta bands, using two EEG datasets from different studies with different oddball paradigms (Study 1: 19 patients with schizophrenia and 17 matched controls, Study 2: 26 patients and 26 controls). Both studies revealed typical P3b ERP components with smaller amplitudes in patients. Reduced ITC in patients was found in the delta band, which correlated with P3b peak amplitudes for all participant groups (ρ = 0.58–0.89). In Study 1, we also found significant differences between patients and controls in ITC in the theta band, which also correlated with P3b peak amplitudes (patients’ ρ = 0.64, controls’ ρ = 0.54). This was not found in Study 2. The results indicate that P3b amplitude reduction in patients with schizophrenia is linked to a reduction in temporal precision of neural activity. These results expand the notion of imprecision in temporal processing at phenomenological, psychological, and neurological levels that have been related to disturbances of the sense of self. They confirm that temporal imprecision may be more important than the reduction of neural activity itself.

Long sequence Hopfield memory*

Abstract Sequence memory is an essential attribute of natural and artificial intelligence that enables agents to encode, store, and retrieve complex sequences of stimuli and actions. Computational models of sequence memory have been proposed where recurrent Hopfield-like neural networks are trained with temporally asymmetric Hebbian rules. However, these networks suffer from limited sequence capacity (maximal length of the stored sequence) due to interference between the memories. Inspired by recent work on Dense Associative Memories, we expand the sequence capacity of these models by introducing a nonlinear interaction term, enhancing separation between the patterns. We derive novel scaling laws for sequence capacity with respect to network size, significantly outperforming existing scaling laws for models based on traditional Hopfield networks, and verify these theoretical results with numerical simulation. Moreover, we introduce a generalized pseudoinverse rule to recall sequences of highly correlated patterns. Finally, we extend this model to store sequences with variable timing between states’ transitions and describe a biologically-plausible implementation, with connections to motor neuroscience.

Distraction by unexpected sounds: comparing response repetition and response switching.

Numerous studies using oddball tasks have shown that unexpected sounds presented in a predictable or repeated sequence (deviant vs. standard sounds) capture attention and negatively impact ongoing behavioral performance. Here, we examine an aspect of this effect that has gone relatively unnoticed: the impact of deviant sounds is stronger for response repetitions than for response switches. Our approach was two-fold. First, we carried out a simulation to estimate the likelihood that stimuli sequences used in past work may not have used balanced proportions of response repetition and switch trials. More specifically, we sought to determine whether the larger distraction effect for response repetitions may have reflected a rarer, and thereby more surprising, occurrence of such trials. To do so, we simulated 10,000 stimuli sets for a 2-AFC task with a proportion of deviant trial of 0.1 or 0.16. Second, we carried out a 2-AFC oddball task in which participants judged the duration of a tone (short vs. long). We carefully controlled the sequence of stimuli to ensure to balance the proportions of response repetitions and response switches across the standard and deviant conditions. The results of the stimuli simulation showed that, contrary to our concerns, response switches were more likely than response repetitions when left uncontrolled for. This suggests that the larger distraction found for response repetition in past work may in fact have been underestimated. In the tone duration judgment task, the results showed a large impact of the response type on distraction as measured by response times: Deviants sounds significantly delayed response repetitions but notably accelerated switches. These findings suggest that deviant sound hinder response repetition and encourage or bias the cognitive system towards a change of responses. We discuss these findings in relation to the adaptive nature of the involuntary detection of unexpected stimuli and in relation to the notion of partial repetition costs. We argue that results are in line with the binding account as well as with the signaling theory.

Using the Cocktail Party Effect to Add the Coding Dimension of Auditory Event Related Potential Brain-Computer Interface.

The auditory event-related potential based brain-computer interface (aERP-BCI) is a classical paradigm of brain-computer communication. To improve the coding efficiency of aERP-BCI, this study proposes a method using two parallel voice channels to add the coding dimension based on the cocktail party effect. The novel paradigm used male and female voices to establish two parallel oddball sound stimulus sequences. In comparison, the baseline paradigm only presented male or female stimulus sequences. Both the double voice condition (DVC) and the single voice condition (SVC) paradigms carried out offline experiments and the DVC also carried out online experiment. Subsequently, the EEG signal and BCI operation results were compared and analyzed. The cocktail party effect caused a significant difference in the EEG responses of non-target stimulus between the focused vocal channel and the ignored vocal channel under the DVC paradigm, and the focused and ignored channels achieved a recognition accuracy of 97.2%. The target recognition rate of DVC was 82.3%, with no significant difference compared with 85% of SVC while the information transfer rate (ITR) of DVC reaching 15.3 bits/min was significantly higher than that of SVC. The cocktail party effect improves the coding efficiency by adding parallel channels without reducing the target/non-target stimulus recognition in the focused vocal channel. This provides a novel direction for the performance improvement of aERP-BCI.

Changes in high-frequency neural inputs to muscles during movement cancellation

Objective.Cortical beta (13-30 Hz) and gamma (30-60 Hz) oscillations are prominent in the motor cortex and are known to be transmitted to the muscles despite their limited direct impact on force modulation. However, we currently lack fundamental knowledge about the saliency of these oscillations at spinal level. Here, we developed an experimental approach to examine the modulations in high-frequency inputs to motoneurons under different motor states while maintaining a stable force, thus constraining behaviour.Approach.Specifically, we acquired brain and muscle activity during a 'GO'/'NO-GO' task. In this experiment, the effector muscle for the task (tibialis anterior) was kept tonically active during the trials, while participants (N= 12) reacted to sequences of auditory stimuli by either keeping the contraction unaltered ('NO-GO' trials), or by quickly performing a ballistic contraction ('GO' trials). Motor unit (MU) firing activity was extracted from high-density surface and intramuscular electromyographic signals, and the changes in its spectral contents in the 'NO-GO' trials were analysed.Main results.We observed an increase in beta and low-gamma (30-45 Hz) activity after the 'NO-GO' cue in the MU population activity. These results were in line with the brain activity changes measured with electroencephalography. These increases in power occur without relevant alterations in force, as behaviour was restricted to a stable force contraction.Significance.We show that modulations in motor cortical beta and gamma rhythms are also present in muscles when subjects cancel a prepared ballistic action while holding a stable contraction in a 'GO'/'NO-GO' task. This occurs while force levels produced by the task effector muscle remain largely unaltered. Our results suggest that muscle recordings are informative also about motor states that are not force-control signals. This opens up new potential use cases of peripheral neural interfaces.

A Coupled Hidden Markov Model framework for measuring the dynamics of categorization

We introduce a new framework for measuring the dynamics of category learning using Coupled Hidden Markov Models (CHMMs). The key assumptions of the framework are that people maintain a latent assignment of every stimulus to a category, and that they can update the assignments for all stimuli whenever they encounter any stimulus. These assumptions contrast with many existing accounts of category learning, which either do not allow for what is learned about one stimulus to influence the category association of others, or allow only for indirect influence. The CHMM framework allows tailored models to be developed for specific category learning tasks, taking as input the stimulus sequence and category responses people make, and producing as output inferences about the underlying dynamics of category assignments and the mechanics of the response processes. We demonstrate the framework by applying it to a categorization task considered by Lee and Navarro (2002), showing how the model measures the change in participants’ latent category assignments as they learn the category structure. We conclude by discussing potential applications of the CHMM framework to category learning situations involving prior knowledge, changing category structures, and category learning tasks that involve the consideration of multiple stimuli at one time.

Combinatorial responsiveness of single chemosensory neurons to external stimulation of mouse explants revealed by DynamicNeuronTracker.

Calcium fluorescence imaging enables us to investigate how individual neurons of live animals encode sensory input or drive specific behaviors. Extracting and interpreting large-scale neuronal activity from imaging data are crucial steps in harnessing this information. A significant challenge arises from uncorrectable tissue deformation, which disrupts the effectiveness of existing neuron segmentation methods. Here, we propose an open-source software, DynamicNeuronTracker (DyNT), which generates dynamic neuron masks for deforming and/or incompletely registered 3D calcium imaging data using patch-matching iterations. We demonstrate that DyNT accurately tracks densely populated neurons, whereas a widely used static segmentation method often produces erroneous masks. DyNT also includes automated statistical analyses for interpreting neuronal responses to multiple sequential stimuli. We applied DyNT to analyze the responses of pheromone-sensing neurons in mice to controlled stimulation. We found that four bile acids and four sulfated steroids activated 15 subpopulations of sensory neurons with distinct combinatorial response profiles, revealing a strong bias toward detecting sulfated estrogen and pregnanolone.

Dynamic, multi-color, multi-level stimulus-responsive luminescence in a co-crystallized coordination polymer: Modulation of energy transfer by acid-induced ligand protonation and photo-induced radical generation

Stimulus-responsive luminescent materials (SLMs), which can respond to external stimuli by switching their luminescence properties, hold great promise for a variety of applications. Particularly, multi-level SLMs with a logical stimulus sequence and a dynamic, multi-color transition are highly desirable for anti-counterfeiting and information encryption, but an effective design strategy is lacking. Herein, a co-crystallized coordination polymer multi-level SLM [Cd(9-AC)4]·(HAD)2·(9-HAC)2 (9-HAC for anthracene-9-carboxylic acid and AD for acridine) with dual emission centers of coordination unit ([Cd(9-AC)4]2-) and exciplex ([9-HAC/HAD]+) has been successfully constructed by a stepwise solvent process. This SLM possesses a variety of stimulus-responsive properties, including mechanochromic (MC), mechanochromic luminescence (MCL), reversible acidochromic luminescence (ACL), and reversible photochromic luminescence (PCL). Notably, both ACL and PCL properties are achieved by modulating the energy transfer process between the coordination unit and the exciplex, while there is a logical sequence between these two stimulus actions. The acid-induced activation and radical-regulated PCL performance has been exploited in the development of anti-counterfeiting and information encryption applications. This work provides an effective strategy to realize dynamic, multi-color, multi-level stimuli-responsive luminescence by modulating the energy transfer through acid-induced ligand protonation and photo-induced radical generation.