Unveiling Adolescent Suicidality: Holistic Analysis of Protective and Risk Factors Using Multiple Machine Learning Algorithms.

Adolescent suicide attempts are on the rise, presenting a significant public health concern. Recent research aimed at improving risk assessment for adolescent suicide attempts has turned to machine learning. But no studies to date have examined the performance of stacked ensemble algorithms, which are more suitable for low-prevalence conditions. The existing machine learning-based research also lacks population-representative samples, overlooks protective factors and their interplay with risk factors, and neglects established theories on suicidal behavior in favor of purely algorithmic risk estimation. The present study overcomes these shortcomings by comparing the performance of a stacked ensemble algorithm with a diverse set of algorithms, performing a holistic item analysis to identify both risk and protective factors on a comprehensive data, and addressing the compatibility of these factors with two competing theories of suicide, namely, The Interpersonal Theory of Suicide and The Strain Theory of Suicide. A population-representative dataset of 173,664 Norwegian adolescents aged 13 to 18 years (mean = 15.14, SD = 1.58, 50.5% female) with a 4.65% rate of reported suicide attempt during the past 12 months was analyzed. Five machine learning algorithms were trained for suicide attempt risk assessment. The stacked ensemble model significantly outperformed other algorithms, achieving equal sensitivity and a specificity of 90.1%, AUC of 96.4%, and AUCPR of 67.5%. All algorithms found recent self-harm to be the most important indicator of adolescent suicide attempt. Exploratory factor analysis suggested five additional risk domains, which we labeled internalizing problems, sleep disturbance, disordered eating, lack of optimism regarding future education and career, and victimization. The identified factors provided stronger support for The Interpersonal Theory of Suicide than for The Strain Theory of Suicide. An enhancement to The Interpersonal Theory based on the risk and protective factors identified by holistic item analysis is presented.

From pre-processing to advanced dynamic modeling of pupil data.

The pupil of the eye provides a rich source of information for cognitive scientists, as it can index a variety of bodily states (e.g., arousal, fatigue) and cognitive processes (e.g., attention, decision-making). As pupillometry becomes a more accessible and popular methodology, researchers have proposed a variety of techniques for analyzing pupil data. Here, we focus on time series-based, signal-to-signal approaches that enable one to relate dynamic changes in pupil size over time with dynamic changes in a stimulus time series, continuous behavioral outcome measures, or other participants' pupil traces. We first introduce pupillometry, its neural underpinnings, and the relation between pupil measurements and other oculomotor behaviors (e.g., blinks, saccades), to stress the importance of understanding what is being measured and what can be inferred from changes in pupillary activity. Next, we discuss possible pre-processing steps, and the contexts in which they may be necessary. Finally, we turn to signal-to-signal analytic techniques, including regression-based approaches, dynamic time-warping, phase clustering, detrended fluctuation analysis, and recurrence quantification analysis. Assumptions of these techniques, and examples of the scientific questions each can address, are outlined, with references to key papers and software packages. Additionally, we provide a detailed code tutorial that steps through the key examples and figures in this paper. Ultimately, we contend that the insights gained from pupillometry are constrained by the analysis techniques used, and that signal-to-signal approaches offer a means to generate novel scientific insights by taking into account understudied spectro-temporal relationships between the pupil signal and other signals of interest.

The lateralized flash-lag illusion: A psychophysical and pupillometry study.

The flash-lag illusion (FLI) is a visual phenomenon where a flashed object, either co-localized or in physical alignment with another continuously moving object, is perceived to lag behind the path of the moving object. In the present study, we reveal an anisotropy of the FLI between the lateral visual fields that was expressed psychophysically as different points of subjective equality, depending on the hemifield in which the stimuli appeared. Specifically, the study confirmed that, as seen in two previous studies, the FLI was significantly larger in the left visual field (LVF) than in the right (RVF). In addition, pupil dilations were larger in the RVF than in the LVF as well as returning to baseline levels more rapidly in the LVF. We interpret these findings as converging on revealing more efficient spatial and attentional processing and, in turn, extrapolation of motion in the LVF/right hemisphere than in the RVF/left hemisphere.

Human voices escape the auditory attentional blink: Evidence from detections and pupil responses.

Attentional selection of a second target in a rapid stream of stimuli embedding two targets tends to be briefly impaired when two targets are presented in close temporal proximity, an effect known as an attentional blink (AB). Two target sounds (T1 and T2) were embedded in a rapid serial auditory presentation of environmental sounds with a short (Lag 3) or long lag (Lag 9). Participants were to first identify T1 (bell or sine tone) and then to detect T2 (present or absent). Individual stimuli had durations of either 30 or 90 ms, and were presented in streams of 20 sounds. The T2 varied in category: human voice, cello, or dog sound. Previous research has introduced pupillometry as a useful marker of the intensity of cognitive processing and attentional allocation in the visual AB paradigm. Results suggest that the interplay of stimulus factors is critical for target detection accuracy and provides support for the hypothesis that the human voice is the least likely to show an auditory AB (in the 90 ms condition). For the other stimuli, accuracy for T2 was significantly worse at Lag 3 than at Lag 9 in the 90 ms condition, suggesting the presence of an auditory AB. When AB occurred (at Lag 3), we observed smaller pupil dilations, time-locked to the onset of T2, compared to Lag 9, reflecting lower attentional processing when 'blinking' during target detection. Taken together, these findings support the conclusion that human voices escape the AB and that the pupillary changes are consistent with the so-called T2 attentional deficit. In addition, we found some indication that salient stimuli like human voices could require a less intense allocation of attention, or noradrenergic potentiation, compared to other auditory stimuli.

More mental rotation time does not imply more mental effort: Pupillary diameters do not change with angular distance.

The ability to mentally rotate objects in space is a fundamental cognitive capacity. Previous studies showed that the time to rotate the image of a figure to match another increases progressively with angular disparity. It remains unclear whether this increase in response time with angular disparity could reflect increased processing operations or more cognitive effort instead of a sustained use of a 'rotate' mechanism without a change in workload. We collected response times as well as pupillary responses that index cognitive workload and activity in the brainstem's locus coeruleus, from a sample of 38 young adults performing a chronometric mental rotations task. The results showed the expected increase in response times but no increase in pupil diameters between 60, 120, and 180 degrees of rotation, suggesting no significant changes in arousal levels when rotating figures near and far. This indicates that during mental rotation the load on cognitive resources remains constant irrespective of angular distance.

'Defrosting' music chills with naltrexone: The role of endogenous opioids for the intensity of musical pleasure.

The endogenous opioid system has been implicated during experiences of pleasure (i.e., from food or sex). Music can elicit intense emotional and bodily sensations of pleasure, called 'Chills'. We investigated the effects of an opioid antagonist (50 mg naltrexone) or placebo (40 µg d3-vitamin) while listening to self-selected music or other 'control' music selected by another participant. We used a novel technique of continuous measurement of pleasantness with an eye tracker system, where participants shifted their eyes along a visual analogue scale, in the semblance of a thermometer so that, as the music unfolded, gaze positions indicated the self-reported hedonic experience. Simultaneously, we obtained pupil diameters. Self-reported pleasure remained unchanged by naltrexone, which - however - selectively decreased pupillary diameters during 'Chills'. Hence, the endogenous µ-opioid signaling is not necessary for subjective enjoyment of music but an opioid blockade dampens pupil responses to peak pleasure, consistent with decreased arousal to the music.

Distinct Neural Mechanisms Meet Challenges in Dynamic Visual Attention due to Either Load or Object Spacing.

When engaged in dynamic visuospatial tasks, the brain copes with perceptual and cognitive processing challenges. During multiple-object tracking (MOT), the number of objects to be tracked (i.e., load) imposes attentional demands, but so does spatial interference from irrelevant objects (i.e., close encounters). Presently, it is not clear whether the effect of load on accuracy solely depends on the number of close encounters. If so, the same cognitive and physiological mechanisms deal with increasing load by preparing for and dealing with spatial interference. However, this has never been directly tested. Such knowledge is important to understand the neurophysiology of dynamic visual attention and resolve conflicting views within visual cognition concerning sources of capacity limitations. We varied the processing challenge in MOT task in two ways: the number of targets and the minimum spatial proximity between targets and distractors. In a first experiment, we measured task-induced pupil dilations and saccades during MOT. In a separate cohort, we measured fMRI activity. In both cohorts, increased load and close encounters (i.e., close spatial proximity) led to reduced accuracy in an additive manner. Load was associated with pupil dilations, whereas close encounters were not. Activity in dorsal attentional areas and frequency of saccades were proportionally larger both with higher levels of load and close encounters. Close encounters recruited additionally ventral attentional areas that may reflect orienting mechanisms. The activity in two brainstem nuclei, ventral tegmental area/substantia nigra and locus coeruleus, showed clearly dissociated patterns. Our results constitute convergent evidence indicating that different mechanisms underlie processing challenges due to load and object spacing.

Pupillary and behavioral markers of alerting and orienting: An individual difference approach.

Measuring task-evoked pupillary (TEP) responses as an index of phasic activity in the locus coeruleus (LC), we examined two competing hypotheses regarding the alerting and orienting mechanisms of attention. According to a dual mechanism account (Fernandez-Duque & Posner, 1997), two separate noradrenergic and cholinergic mechanisms modulate, respectively, the alerting and orienting effects. However, Corbetta and colleagues (2008) proposed that LC phasic activity may also be involved in orienting effect through its functional relationship with the ventral attentional network. We recruited seventy-five healthy Norwegian participants to perform a Posner cueing task. Both behavioral and pupillary responses revealed the alerting effect. Also, both behavioral and pupillary responses indicated that cued attention is affected by age. Behavioral responses also revealed orienting effect However, we found no TEP differences between valid, invalid, and neutral conditions, suggesting that TEP effects were driven by the alerting effect of cue presentation. Moreover, both behavioral and pupillary estimates of alertness and orienting were uncorrelated. Finally, individual differences in general cognitive abilities did not appear to affect the orienting and alerting mechanisms. This pattern of results is consistent with the dual mechanism account of attention. However, the LC involvement in the (re)orienting attention may be driven by state-specific factors.

Cueing the Necker cube: Pupil dilation reflects the viewing-from-above constraint in bistable perception.

We hypothesized that a perceptually ambiguous or bistable object (Necker cube) can be more effectively biased to assume a point of view-from-above (VFA) than from below the object by cueing attention. Participants viewed a Necker cube in which one surface was temporarily shaded so as to prime a specific perspective on the cube. Subsequently, the standard (wireframe) Necker cube was viewed for 3 seconds, and participants reported what perspective they had seen initially and whether their perception shifted to the alternative perspective during the brief viewing. Concomitantly, pupil size was monitored with an eye-tracker to obtain an index of cognitive effort. There were two conditions: passive viewing and forced attention to sustain the initially primed perspective. We confirmed the presence of a VFA bias with forced attention, which was accompanied by reduced attentional effort, as indexed by a reduced pupil diameter, compared with the view-from-below. Participants showed no bias during passive viewing. We suggest that the level of intensive attention, when retrieving and maintaining a specific view from memory, is mirrored in the size of the eye pupils and may reflect ecological constraints on visual perception.

Ocular signatures of proactive versus reactive cognitive control in young adults.

During the execution of a cognitive task, the brain maintains contextual information to guide behavior and achieve desired goals. The AX-Continuous Performance Task is used to study proactive versus reactive cognitive control. Young adults tend to behave proactively in standard testing conditions. However, it remains unclear how interindividual variability (e.g., in cognitive and motivational factors) may drive people into more reactive or proactive control under the same task demands. We investigated the use of control strategies in a large population of healthy young adults. We computed the proactive behavioral index and consequently divided participants into proactive, reactive, and intermediate groups. We found that reactive participants were generally slower, presented lower context sensitivity, and larger response variability. Pupillary changes and blink rate index cognitive effort allocation. We measured, concomitantly to the task, the pupil size and frequency of blinks associated with the cue maintenance and response intervals. During the cue period, nonfrequent, nontarget cues led to increased pupil dilation and number of blinks in all participants. During the response interval, we found more errors and increased pupil dilation to the probe when all participants had to overcome a response bias generated by the frequent cue. Only reactive participants showed larger response-related pupil when they had to overcome a response bias related to the frequent probe. Contrary to expectations, groups did not differ in ocular measures in the cue period. In conclusion, interindividual differences in cognitive control between healthy adults can be mapped onto different patterns of effort allocation indexed by the pupil.

Modulating adaptation to emotional faces by spatial frequency filtering.

In four experiments, we investigated the presence and strength of perceptual aftereffects to emotional faces, using spatial frequency filtering to manipulate awareness of emotional content. We presented angry and happy faces as adapters and used a control condition without adaptation. Participants were subsequently requested to judge the friendliness level of a neutral target face. We confirmed the well-known aftereffect for unfiltered emotional faces in Experiment 1. In the experiment, friendliness judgments were greater for the angry than the happy or the control condition. In Experiment 2, in which the "hidden" emotional expression contained in the low spatial frequencies (LSF) was superimposed to the neutral expression presented in the rest of SF of the same image (emotional hybrid faces), the difference in friendliness judgments between angry and happy was significant, but none of the two conditions significantly differed from the control. In Experiment 3, faces were presented at LSF, confirmed a difference between the two emotions, but only the judgments of angry faces differed from the control condition. In Experiment 4, we extended the initial finding to stimuli presented at middle and high spatial frequencies (M-HSF). Finally, a comparison among all experiments revealed that the aftereffect was stronger with angry faces filtered at M-HSF than all of the other filtering conditions, whereas there were no differences for happy faces. We conclude that spatial frequency filtering influences aftereffects and that these effects are also related to emotional awareness. The results are discussed with reference to the dual route model of visual perception.

Binocular rivalry after right-hemisphere stroke: Effects of attention impairment on perceptual dominance patterns.

Binocular rivalry is when perception fluctuates while the stimuli, consisting of different images presented to each eye, remain unchanged. The fluctuation rate and predominance ratio of these images are regarded as information source for understanding properties of consciousness and perception. We administered a binocular rivalry task to 26 right-hemisphere stroke patients and 26 healthy control participants, using stimuli such as simple Gabor anaglyphs. Each single Gabor image was of unequal spatial frequency compared to its counterpart, allowing assessment of the effect of relative spatial frequency on rivalry predominance. Results revealed that patients had significantly decreased alternation rate compared to healthy controls, with severity of patients' attention impairment predicting alternation rates. The patient group had higher predominance ratio for high compared to low relative spatial frequency stimuli consistent with the hypothesis that damage to the right hemisphere may disrupt processing of relatively low spatial frequencies. Degree of attention impairment also predicted the effect of relative spatial frequencies. Lastly, both groups showed increased predominance rates in the right eye compared to the left eye. This right eye dominance was more pronounced in patients than controls, suggesting that right hemisphere stroke may additionally affect eye predominance ratios.

Using space to represent categories: insights from gaze position.

We investigated the boundaries among imagery, memory, and perception by measuring gaze during retrieved versus imagined visual information. Eye fixations during recall were bound to the location at which a specific stimulus was encoded. However, eye position information generalized to novel objects of the same category that had not been seen before. For example, encoding an image of a dog in a specific location enhanced the likelihood of looking at the same location during subsequent mental imagery of other mammals. The results suggest that eye movements can also be launched by abstract representations of categories and not exclusively by a single episode or a specific visual exemplar.

Music chills: The eye pupil as a mirror to music's soul.

This study evaluated whether music-induced aesthetic "chill" responses, which typically correspond to peak emotional experiences, can be objectively monitored by degree of pupillary dilation. Participants listened to self-chosen songs versus control songs chosen by other participants. The experiment included an active condition where participants made key presses to indicate when experiencing chills and a passive condition (without key presses). Chills were reported more frequently for self-selected songs than control songs. Pupil diameter was concurrently measured by an eye-tracker while participants listened to each of the songs. Pupil size was larger within specific time-windows around the chill events, as monitored by key responses, than in comparison to pupil size observed during 'passive' song listening. In addition, there was a clear relationship between pupil diameter within the chills-related time-windows during both active and passive conditions, thus ruling out the possibility that chills-related pupil dilations were an artifact of making a manual response. These findings strongly suggest that music chills can be visible in the moment-to-moment changes in the size of pupillary responses and that a neuromodulatory role of the central norepinephrine system is thereby implicated in this phenomenon.

Why Saturday could be both green and red in synesthesia.

It has long been observed that certain words induce multiple synesthetic colors, a phenomenon that has remained largely unexplored. We report here on the distinct synesthetic colors two synesthetes experienced with closed sets of concepts (digits, weekdays, months). For example, Saturday was associated with green, like other word starting with s; however, Saturday also had its specific color (red). Auditory priming and Visual Color Stroop tasks were used to understand the cognitive mechanisms supporting the distinct synesthetic colors. Results revealed that processing of word segments and whole words was specifically involved in each type of synesthetic colors. However, these mechanisms differed between participants, as they could relate either to orthography (and written words) or phonology (and spoken words). Further differences concerned the word representations, which varied as to whether or not they encoded serial positions. In addition to clarifying the cognitive mechanisms underlying the distinct synesthetic colors, our results offer some clues for understanding the neurocognitive underpinnings of a rather common form of synesthesia.

Functional connectivity indicates differential roles for the intraparietal sulcus and the superior parietal lobule in multiple object tracking.

Attentive tracking requires sustained object-based attention, rather than passive vigilance or rapid attentional shifts to brief events. Several theories of tracking suggest a mechanism of indexing objects that allows for attentional resources to be directed toward the moving targets. Imaging studies have shown that cortical areas belonging to the dorsal frontoparietal attention network increase BOLD-signal during multiple object tracking (MOT). Among these areas, some studies have assigned IPS a particular role in object indexing, but the neuroimaging evidence has been sparse. In the present study, we tested participants on a continuous version of the MOT task in order to investigate how cortical areas engage in functional networks during attentional tracking. Specifically, we analyzed the data using eigenvector centrality mapping (ECM) analysis, which provides estimates of individual voxels' connectedness with hub-like parts of the functional network. The results obtained using permutation based voxel-wise statistics support the proposed role for the IPS in object indexing as this region displayed increased centrality during tracking as well as increased functional connectivity with both prefrontal and visual perceptual cortices. In contrast, the opposite pattern was observed for the SPL, with decreasing centrality, as well as reduced functional connectivity with the visual and frontal cortices, in agreement with a hypothesized role for SPL in attentional shifts. These findings provide novel evidence that IPS and SPL serve different functional roles during MOT, while at the same time being highly engaged during tracking as measured by BOLD-signal changes.

The cerebral correlates of subliminal emotions: an eleoencephalographic study with emotional hybrid faces.

In a high-resolution electroencephalographic study, participants evaluated the friendliness level of upright and inverted 'hybrid faces', i.e. facial photos containing a subliminal emotional core in the low spatial frequencies (< 6 cycles/image), superimposed on a neutral expression in the rest of the spatial frequencies. Upright happy and angry faces were judged as more friendly or less friendly than neutral faces, respectively. We observed the time course of cerebral correlates of these stimuli with event-related potentials (ERPs), confirming that hybrid faces elicited the posterior emotion-related and face-related components (P1, N170 and P2), previously shown to be engaged by non-subliminal emotional stimuli. In addition, these components were stronger in the right hemisphere and were both enhanced and delayed by face inversion. A frontal positivity (210-300 ms) was stronger for emotional than for neutral faces, and for upright than for inverted faces. Hence, hybrid faces represent an original approach in the study of subliminal emotions, which appears promising for investigating their electrophysiological correlates.

Bright illusions reduce the eye's pupil.

We recorded by use of an infrared eye-tracker the pupil diameters of participants while they observed visual illusions of lightness or brightness. Four original illusions {based on Gaetano Kanisza's [Kanizsa G (1976) Subjective contours. Sci Am 234:48-52] and Akiyoshi Kitaoka's [Kitaoka A. (2005) Trick Eyes (Barnes & Noble, New Providence, NJ).] examples} were manipulated to obtain control conditions in which the perceived illusory luminance was either eliminated or reduced. All stimuli were equiluminant so that constrictions in pupillary size could not be ascribed to changes in light energy. We found that the pupillary diameter rapidly varied according to perceived brightness and lightness strength. Differences in local contrast information could be ruled out as an explanation because, in a second experiment, the observers maintained eye fixation in the center of the display; thus, differential stimulation of the fovea by local contrast changes could not be responsible for the pupillary differences. Hence, the most parsimonious explanation for the present findings is that pupillary responses to ambient light reflect the perceived brightness or lightness of the scene and not simply the amount of physical light energy entering the eye. Thus, the pupillary physiological response reflects the subjective perception of light and supports the idea that the brain's visual circuitry is shaped by visual experience with images and their possible sources.

Tracking down the path of memory: eye scanpaths facilitate retrieval of visuospatial information.

Recent research points to a crucial role of eye fixations on the same spatial locations where an item appeared when learned, for the successful retrieval of stored information (e.g., Laeng et al. in Cognition 131:263-283, 2014. doi: 10.1016/j.cognition.2014.01.003 ). However, evidence about whether the specific temporal sequence (i.e., scanpath) of these eye fixations is also relevant for the accuracy of memory remains unclear. In the current study, eye fixations were recorded while looking at a checkerboard-like pattern. In a recognition session (48 h later), animations were shown where each square that formed the pattern was presented one by one, either according to the same, idiosyncratic, temporal sequence in which they were originally viewed by each participant or in a shuffled sequence although the squares were, in both conditions, always in their correct positions. Afterward, participants judged whether they had seen the same pattern before or not. Showing the elements serially according to the original scanpath's sequence yielded a significantly better recognition performance than the shuffled condition. In a forced fixation condition, where the gaze was maintained on the center of the screen, the advantage of memory accuracy for same versus shuffled scanpaths disappeared. Concluding, gaze scanpaths (i.e., the order of fixations and not simply their positions) are functional to visual memory and physical reenacting of the original, embodied, perception can facilitate retrieval.

The eye pupil adjusts to imaginary light.

If a mental image is a rerepresentation of a perception, then properties such as luminance or brightness should also be conjured up in the image. We monitored pupil diameters with an infrared eye tracker while participants first saw and then generated mental images of shapes that varied in luminance or complexity, while looking at an empty gray background. Participants also imagined familiar scenarios (e.g., a "sunny sky" or a "dark room") while looking at the same neutral screen. In all experiments, participants' eye pupils dilated or constricted, respectively, in response to dark and bright imagined objects and scenarios. Shape complexity increased mental effort and pupillary sizes independently of shapes' luminance. Because the participants were unable to voluntarily constrict their eyes' pupils, the observed pupillary adjustments to imaginary light present a strong case for accounts of mental imagery as a process based on brain states similar to those that arise during perception.