AI is already playing a transformative role in the field. Scientists are leveraging powerful AI models to analyse huge neural datasets and to create predictive models that mirror the brain’s own processes.1 This fundamental understanding is the bedrock for ambitious goals like building functional digital twins of specific brain regions,2 which could dramatically accelerate the development of therapies.
This synergy is a two-way street: insights into the brain’s remarkable energy efficiency are inspiring the development of next-generation, brain-like "neuromorphic" computer chips, promising to make powerful AI accessible on small, everyday devices.3
The immediate focus is on developing sophisticated closed-loop brain-computer interfaces (BCIs) to treat conditions like depression by reading and actively regulating brain states in real time.4 This requires overcoming significant hurdles in materials science to create biocompatible, flexible electronics that can integrate safely and effectively with neural tissue.
Eventually, these technologies will extend our minds — both in cognitive capabilities and in the physical location of data-logging and processing. Invasive implants, such as those restoring motor function, and non-invasive wearables are already offloading and augmenting cognitive tasks. We are likely to become ever more reliant on such tools, and research will make their integration into our lives ever more seamless.
However impressive the advances, progress in this area also creates obvious ethical issues about permission to read from and write to the brain, ownership of neural data and inequities emerging in cognitive ability, to name just a few concerns. Navigating the landscape of data privacy, technological dependency and the integration of AI with human consciousness will be as critical as the scientific breakthroughs themselves.
KEY TAKEAWAYS
The convergence of neuroscience, sensing technology and materials science is paving the way for unprecedented interventions in understanding, augmenting and repairing the functions of the human brain. Efforts to understand the Fundamentals of cognition are progressing through a range of technologies that can generate and analyse huge neural datasets. Researchers are also making progress in Decrypting the brain, using AI to understand how neural activity leads to thought, intention and behaviour. This is allowing the construction of useful Neuromodulation systems: precise, targeted interventions, often involving feedback loops, that can influence attention, learning, memory and affective states. As a result, we are moving into an era of Exogenous cognition that will integrate artificial cognitive systems with biological brains. Such developments are set to fundamentally alter human experience.
Three fundamental questions guide GESDA’s mission and drive its work: Who are we, as humans? How can we all live together? How can we ensure the well-being of humankind and the sustainable future of our planet? We asked researchers from the field to anticipate what impact future breakthroughs could have on each of these dimensions. This wheel summarises their opinions when considering each of these questions, with a higher score indicating high anticipated impact, and vice versa.
It is now clear that cognitive processes are dynamic and distributed, making them extremely difficult not only to investigate but also to model.6 However, researchers are beginning to unify insights across species and scales — from synaptic plasticity to network-level population dynamics — and are moving beyond investigating isolated brain regions toward an understanding of integrated, systems-level frameworks.7
AI is also beginning to play an important role in developing our understanding of the brain. Researchers are using large-scale AI models to analyse massive neural datasets and to identify functions of neural systems. Deep learning can be trained to predict neural responses and develop representations similar to those in the biological brain. This approach is helping to identify the underlying mechanisms of cognition.8 Meanwhile, generative AI systems are able to reconstruct images and speech directly from brain activity, offering a potential window into internal thought processes and new avenues for restoring communication.
A more ambitious goal is the creation of a functional digital twin of the brain.9 This would be a high-fidelity computational model focused on the brain’s information processing rather than its precise physical structure. Achieving this requires the analysis and fusion of high-resolution neural data from many experiments and individuals to build a predictive model capable of running in silico experiments on brain behaviour. Achieving this could dramatically accelerate the research and development of neural therapies.
A better understanding of human cognition, which is extremely data- and energy-efficient, could also lead to more efficient AI systems built by mimicking the brain’s architecture. Such “neuromorphic” computing systems are being developed for two key areas. The first is heterogeneous computing, where energy-efficient neuromorphic chips handle specific tasks within conventional devices. The second, and potentially more impactful, is “edge” computing, enabling powerful AI to run on small, portable devices without relying on the cloud.10
The Anticipation Potential of a research field is determined by the capacity for impactful action in the present, considering possible future transformative breakthroughs in a field over a 25-year outlook. A field with a high Anticipation Potential, therefore, combines the potential range of future transformative possibilities engendered by a research area with a wide field of opportunities for action in the present. We asked researchers in the field to anticipate:
This chart represents a summary of their responses to each of these elements, which when combined, provide the Anticipation Potential for the topic. See methodology for more information.
Progress in decrypting neural states will provide opportunities to regulate brain states using brain-computer interfaces (BCIs). The use of AI in the development of BCIs has potential for revolutionising treatments for neuropsychiatric disorders and for the possibility of enhancing human cognition.16 Disentangling the neural signals relevant to a specific condition, like depression, from the myriad of other brain activities remains a challenge. The vision is to develop sophisticated closed-loop systems that would continuously record brain activity while AI algorithms decode internal states such as mood and then guide personalised therapies in real time. These interventions could range from deep brain stimulation to pharmacology.
A more futuristic and complex vision involves a true fusion of the brain and computer, creating a single, enhanced mind. However, a fundamental obstacle is the unique variability of individual brains, which can be modelled only on a neuron-by-neuron scale. The scale of this challenge suggests that this kind of model of the brain will remain out of reach for the foreseeable future.
The Anticipation Potential of a research field is determined by the capacity for impactful action in the present, considering possible future transformative breakthroughs in a field over a 25-year outlook. A field with a high Anticipation Potential, therefore, combines the potential range of future transformative possibilities engendered by a research area with a wide field of opportunities for action in the present. We asked researchers in the field to anticipate:
This chart represents a summary of their responses to each of these elements, which when combined, provide the Anticipation Potential for the topic. See methodology for more information.
Their operation relies on understanding — gained through techniques ranging from deep-brain stimulation and transcranial electrical and magnetic stimulation to optogenetics and chemogenetics — of how specific patterns of stimulation can shape behaviour and cognition. Yet the future of neuromodulation lies not only in stimulation techniques, but in adaptive, feedback-driven systems that can respond to neural and behavioural states in real time.17
Materials science remains a significant roadblock. Current implantable systems are built from a narrow range of materials, usually dry, rigid electronics, that are in stark contrast to biological tissue, which is soft and dynamic. This disparity hinders the development of higher-resolution brain-computer interfaces: future progress depends on materials that are flexible enough to cope with the movement of neural tissue, biocompatible and capable of being miniaturised and packaged in a way that protects the electronics within the body.18 While current research is promising, introducing new materials and manufacturing techniques is a slow process because of the extensive testing and approval required.19
Beyond electronics, alternative forms of modulation include optogenetics20 and sonogenetics,21 which offer higher selectivity but require substantial material and manufacturing innovations. Significant clinical testing will also be necessary.
Another key area of development is the detailed mapping of the brain’s connectome, cell function and spatial arrangement on the scale of individual neurons.22 These maps are already becoming available for animal models and will enable more precise, targeted neuromodulation and closed-loop systems that can record and stimulate neural activity with high precision.
The Anticipation Potential of a research field is determined by the capacity for impactful action in the present, considering possible future transformative breakthroughs in a field over a 25-year outlook. A field with a high Anticipation Potential, therefore, combines the potential range of future transformative possibilities engendered by a research area with a wide field of opportunities for action in the present. We asked researchers in the field to anticipate:
This chart represents a summary of their responses to each of these elements, which when combined, provide the Anticipation Potential for the topic. See methodology for more information.
As brain-computer interfaces mature, and as models become more context-aware and biologically informed, new hybrid cognitive architectures may emerge. Non-invasive BCIs, such as glasses23 and earbuds with integrated physiological sensors, are already collecting vast amounts of data,24 which, when combined with AI, could lead to cognitive offloading and augmentation.25,26
Invasive BCIs — implants inserted into the brain via a small craniotomy27 — have even deeper potential. Neuralink, a privately funded company, has developed a hermetically sealed implant that translates motor intentions into digital commands (and vice versa) with immediate applications in restoring function for individuals with paralysis. Patients are already using Neuralink interfaces to control a virtual hand and play games, and the ambition is significantly greater: the company envisions an ability to treat blindness and aphasia, the full-body control of exoskeletons, and perhaps pain and mood management.
Such advanced BCIs raise significant concerns about data privacy,28 the potential for misuse of personal brain data and the risk of creating a dependency on technology that could diminish natural cognitive abilities. Given the eventual (beyond 2050) possibilities of direct cerebral advertisements, identity hacking and the potential for identity destruction, the ethical landscape that will emerge over the next few decades will be complex. A careful, nuanced and informed approach will be required as AI integrates with human consciousness.
The Anticipation Potential of a research field is determined by the capacity for impactful action in the present, considering possible future transformative breakthroughs in a field over a 25-year outlook. A field with a high Anticipation Potential, therefore, combines the potential range of future transformative possibilities engendered by a research area with a wide field of opportunities for action in the present. We asked researchers in the field to anticipate:
This chart represents a summary of their responses to each of these elements, which when combined, provide the Anticipation Potential for the topic. See methodology for more information.
