Brain-machine convergent evolution: Why finding parallels between brain and artificial systems is informative.

Central nervous system neurons manifest a rich diversity of selectivity profiles-whose precise role is still poorly understood. Following the striking success of artificial networks, a major debate has emerged concerning their usefulness in explaining neuronal properties. Here we propose that finding parallels between artificial and neuronal networks is informative precisely because these systems are so different from each other. Our argument is based on an extension of the concept of convergent evolution-well established in biology-to the domain of artificial systems. Applying this concept to different areas and levels of the cortical hierarchy can be a powerful tool for elucidating the functional role of well-known cortical selectivities. Importantly, we further demonstrate that such parallels can uncover novel functionalities by showing that grid cells in the entorhinal cortex can be modeled to function as a set of basis functions in a lossy representation such as the well-known JPEG compression. Thus, contrary to common intuition, here we illustrate that finding parallels with artificial systems provides novel and informative insights, particularly in those cases that are far removed from realistic brain biology.

Face-Selective Units in Human Ventral Temporal Cortex Reactivate during Free Recall.

Research in functional neuroimaging has suggested that category-selective regions of visual cortex, including the ventral temporal cortex (VTC), can be reactivated endogenously through imagery and recall. Face representation in the monkey face-patch system has been well studied and is an attractive domain in which to explore these processes in humans. The VTCs of 8 human subjects (4 female) undergoing invasive monitoring for epilepsy surgery were implanted with microelectrodes. Most (26 of 33) category-selective units showed specificity for face stimuli. Different face exemplars evoked consistent and discriminable responses in the population of units sampled. During free recall, face-selective units preferentially reactivated in the absence of visual stimulation during a 2 s window preceding face recall events. Furthermore, we show that in at least 1 subject, the identity of the recalled face could be predicted by comparing activity preceding recall events to activity evoked by visual stimulation. We show that face-selective units in the human VTC are reactivated endogenously, and present initial evidence that consistent representations of individual face exemplars are specifically reactivated in this manner.SIGNIFICANCE STATEMENT The role of "top-down" endogenous reactivation of native representations in higher sensory areas is poorly understood in humans. We conducted the first detailed single-unit survey of ventral temporal cortex (VTC) in human subjects, showing that, similarly to nonhuman primates, humans encode different faces using different rate codes. Then, we demonstrated that, when subjects recalled and imagined a given face, VTC neurons reactivated with the same rate codes as when subjects initially viewed that face. This suggests that the VTC units not only carry durable representations of faces, but that those representations can be endogenously reactivated via "top-down" mechanisms.

Where Does Time Go When You Blink?

Retinal input is frequently lost because of eye blinks, yet humans rarely notice these gaps in visual input. Although previous studies focused on the perceptual and neural correlates of diminished awareness to blinks, the impact of these correlates on the perceived time of concurrent events is unknown. Here, we investigated whether the subjective sense of time is altered by spontaneous blinks. We found that participants (N = 22) significantly underestimated the duration of a visual stimulus when a spontaneous blink occurred during stimulus presentation and that this underestimation was correlated with the blink duration of individual participants. Importantly, the effect was not present when durations of an auditory stimulus were judged (N = 23). The results point to a link between spontaneous blinks, previously demonstrated to induce activity suppression in the visual cortex, and a compression of subjective time. They suggest that ongoing encoding within modality-specific sensory cortices, independent of conscious awareness, informs the subjective sense of time.

Covert neurofeedback without awareness shapes cortical network spontaneous connectivity.

Recent advances in blood oxygen level-dependent-functional MRI (BOLD-fMRI)-based neurofeedback reveal that participants can modulate neuronal properties. However, it is unknown whether such training effects can be introduced in the absence of participants' awareness that they are being trained. Here, we show unconscious neurofeedback training, which consequently produced changes in functional connectivity, introduced in participants who received positive and negative rewards that were covertly coupled to activity in two category-selective visual cortex regions. The results indicate that brain networks can be modified even in the complete absence of intention and awareness of the learning situation, raising intriguing possibilities for clinical interventions.

Quenching of spontaneous fluctuations by attention in human visual cortex.

In the absence of a task, the human brain enters a mode of slow spontaneous fluctuations. A fundamental, unresolved question is whether these fluctuations are ongoing and thus persist during task engagement, or alternatively, are quenched and replaced by task-related activations. Here, we examined this issue in the human visual cortex, using fMRI. Participants were asked to either perform a recognition task of randomly appearing face and non-face targets (attended condition) or watch them passively (unattended condition). Importantly, in approximately half of the trials, all sensory stimuli were absent. Our results show that even in the absence of stimuli, spontaneous fluctuations were suppressed by attention. The effect occurred in early visual cortex as well as in fronto-parietal attention network regions. During unattended trials, the activity fluctuations were negatively linked to pupil diameter, arguing against attentional fluctuations as underlying the effect. The results demonstrate that spontaneous fluctuations do not remain unchanged with task performance, but are rather modulated according to behavioral and cognitive demands.

Convergent evolution of face spaces across human face-selective neuronal groups and deep convolutional networks.

The discovery that deep convolutional neural networks (DCNNs) achieve human performance in realistic tasks offers fresh opportunities for linking neuronal tuning properties to such tasks. Here we show that the face-space geometry, revealed through pair-wise activation similarities of face-selective neuronal groups recorded intracranially in 33 patients, significantly matches that of a DCNN having human-level face recognition capabilities. This convergent evolution of pattern similarities across biological and artificial networks highlights the significance of face-space geometry in face perception. Furthermore, the nature of the neuronal to DCNN match suggests a role of human face areas in pictorial aspects of face perception. First, the match was confined to intermediate DCNN layers. Second, presenting identity-preserving image manipulations to the DCNN abolished its correlation to neuronal responses. Finally, DCNN units matching human neuronal group tuning displayed view-point selective receptive fields. Our results demonstrate the importance of face-space geometry in the pictorial aspects of human face perception.

The Noisy Brain: Power of Resting-State Fluctuations Predicts Individual Recognition Performance.

The unique profile of strong and weak cognitive traits characterizing each individual is of a fundamental significance, yet their neurophysiological underpinnings remain elusive. Here, we present intracranial electroencephalogram (iEEG) measurements in humans pointing to resting-state cortical "noise" as a possible neurophysiological trait that limits visual recognition capacity. We show that amplitudes of slow (<1 Hz) spontaneous fluctuations in high-frequency power measured during rest were predictive of the patients' performance in a visual recognition 1-back task (26 patients, total of 1,389 bipolar contacts pairs). Importantly, the effect was selective only to task-related cortical sites. The prediction was significant even across long (mean distance 4.6 ± 2.8 days) lags. These findings highlight the level of the individuals' internal "noise" as a trait that limits performance in externally oriented demanding tasks.