Understanding how human decision-making and preferences manifest before conscious thought has long challenged researchers focused on cognitive and information science. Now, the field of neuromarketing – a discipline that looks at the neurocognitive underpinnings of consumer behavior – is starting to uncover, in amazing detail, exactly how the brain goes about recognizing a brand. An international research team based in Auckland University of Technology and Nottingham Trent University has devised a new machine learning method that tracks brain responses to logos on the millisecond timescale…even before conscious thoughts are formed. Their results shed light on the early spikes in brain activity that are tied to brand awareness.
The method utilizes one of the most promising recent trends in artificial intelligence research: spiking neural networks. These networks use algorithms loosely modeled on the behavior of the human brain to recognize patterns in sets of streaming data. The team’s approach can decipher patterns of brain activity occurring at the sub-conscious level following exposure to different stimuli.
To test the model’s power, the researchers collected electroencephalography data from 20 individuals shown familiar and unfamiliar stimuli in the form of marketing logos. EEG provides a measure of brain activity with millisecond precision, making it possible to investigate early cognitive changes, including those occurring pre-consciously. With the model, the team mapped EEG data to a Spiking Neural Network, which used a three-dimensional template to identify regional brain activity. Using this information, the model learned and classified patterns in brain responses. This allowed the researchers to follow the firing of different neurons over time, enabling them to track pre-conscious brain activity.
The model showed that familiar logos elicited more widespread brain responses than unfamiliar logos. The differences were most prominent around 200 milliseconds, but variations in brain activity could be picked up as early as 100 milliseconds after a familiar logo was shown. Insights into this early window of neurocognitive processing have been technically infeasible so far.
Although the model was used to look at the brain activity tied to brand awareness, the methodology holds promise for early detection of brain responses to all types of stimuli. Future applications of this approach may help uncover the cognitive foundations underlying complex processes such as decision-making and learning.