A recent study published in Nature Communications offers new insights into one of the most fundamental aspects of human cognition — our ability to recognize faces. The study, conducted by a team of scientists using intracerebral electrodes to obtain single neuron recordings, provides a unique glimpse into the intricacies of neural responses in the midfusiform gyrus.
Our ability to recognize and distinguish between faces is crucial for social interactions and communication. Previous research has shown that a region in the brain known as the midfusiform gyrus plays a vital role in this process. This area has been linked to the recognition of familiar faces, such as those of family members and friends, as well as famous individuals.
However, the exact workings of the neurons in the midfusiform gyrus have remained elusive. This region’s involvement in face recognition has been observed through techniques like functional magnetic resonance imaging (fMRI), but researchers lacked direct access to individual neurons in this area. To address this gap in knowledge, a team of scientists embarked on a study that involved recording the activity of neurons in the midfusiform gyrus, aiming to unravel the neural basis of familiarity recognition.
“For many years, I have worked with single neuron recordings in the human hippocampus,” said study author Rodrigo Quian Quiroga, an ICREA research professor at the Hospital del Mar Research Institute in Barcelona and author of “NeuroScience Fiction.”
“These recordings are done in epileptic patients for clinical reasons and due to the known involvement of the hippocampus in memory, I have studied memory processes. In Nancy, France, for clinical reasons they also implant electrodes in the fusiform gyrus, and with the team of my colleague, Bruno Rossion, an expert in face perception, we therefore set out to study the encoding of faces in the area that is known (through imaging studies) to be involved in recognizing faces.”
The research team conducted their study with the participation of five patients who had previously undergone surgery for treatment-resistant epilepsy. These patients had intracerebral microelectrodes implanted in their brains as part of their clinical treatment. These electrodes allowed the researchers to directly record the activity of individual neurons in the midfusiform gyrus while the patients viewed various images, providing a unique opportunity for the researchers to gain insights into neural responses that are typically inaccessible.
During the experiments, the patients were shown pictures falling into four categories: familiar faces, unknown faces, familiar places, and unknown places. The researchers aimed to uncover how neurons in this brain region responded to these different visual stimuli. Importantly, the electrodes used for single-neuron recordings were confirmed to be placed in the midfusiform gyrus through independent experiments involving fMRI and Local Field Potential (LFP) face localizers.
One of the most significant findings of this study was the diverse responses exhibited by midfusiform neurons. Out of the 78 units recorded (comprising 33 single units and 45 multi-units), 46% responded to faces, 30% responded to places, and 24% responded to both. This indicates that the midfusiform gyrus is not exclusively dedicated to processing faces but also plays a role in processing place-related information.
“After lot of work setting up the single neuron recordings in the human midfusiform gyrus, it was a thrill to see many neurons responding very strongly and selectively to faces,” Quiroga told PsyPost.
Additionally, the study examined whether neurons in this region exhibited differential responses to familiar and unknown stimuli. Surprisingly, the researchers found that there were no significant differences in the strength, latency, or selectivity of responses between familiar and unknown faces or places.
In other words, these neurons responded similarly to both familiar and unfamiliar stimuli. This challenges previous assumptions about the brain’s role in processing familiar faces and suggests that the midfusiform gyrus is not exclusively tuned to recognize familiar individuals.
While the average responses did not differ significantly, a deeper analysis using single-trial population decoding revealed an intriguing pattern. Despite the lack of pronounced average differences, the collective activity of midfusiform neurons enabled the reliable discrimination of familiar and unknown faces at the population level. This means that, taken together, the neurons in this region contain information about the familiarity of individual faces, even if this information is not overtly apparent in individual neuron responses.
The researchers further identified three distinct types of face-related discriminations occurring at the population level within the midfusiform gyrus. First, there was face detection, which involved distinguishing between faces and non-face objects. Second, familiarity face recognition, which entailed discriminating familiar faces from unknown ones. And third, picture identification, which involved recognizing individual face and place pictures. Each of these discriminations had a different time profile, with face detection occurring earliest in the processing sequence.
“We show the first basic principles of how single neurons encode faces, with the majority of neurons responding selectively to faces (and other neurons responding to other category of stimuli, like places),” Quiroga said. “We also show that at the single neuron level we could see a clear discrimination of face familiarity, something that was not clear from imaging studies.”
To gain a better understanding of how the midfusiform gyrus fits into the larger neural landscape, the researchers compared the properties of these neurons with those found in the human hippocampus, a region crucial for memory. Notably, midfusiform neurons demonstrated lower baseline activity, more graded tuning, and a higher responsiveness to stimuli compared to the more binary responses of hippocampal neurons, which primarily respond to familiar stimuli.
While this study has provided invaluable insights into the workings of the midfusiform gyrus, there are certain limitations to consider. The number of participants in the study was relatively small, involving only five patients. Additionally, the placement of electrodes varied among patients, potentially affecting the results. Therefore, future research with larger and more standardized participant groups may help confirm and expand upon these findings.
“We still need to sort out the underlying code by which these neurons are involved in face recognition,” Quiroga explained. “The caveat is that these recordings and quite rare and we need much more data to tackle this question. We are working on it.”
The researcher added that this study is the “first of what I hope it will be a series of groundbreaking studies showing how faces are coded at the single neuron level in humans and how this compares to what has been described in monkeys (in my hippocampal recordings, I have found what I believe are unique coding features in the human hippocampus; i.e. something that does not exist in other species – whether this is the same or not in the fusiform gyrus is an open question).”
The study, “Single neuron responses underlying face recognition in the human midfusiform face-selective cortex“, was authored by Rodrigo Quian Quiroga, Marta Boscaglia, Jacques Jonas, Hernan G. Rey, Xiaoqian Yan, Louis Maillard, Sophie Colnat-Coulbois, Laurent Koessler, and Bruno Rossion.
