Abstract
Aging is accompanied by unisensory decline. To compensate for this, older adults may rely increasingly on two strategies: integrating more information from different sensory organs and utilizing internal "templates" or "priors" of the environment, as suggested by the predictive coding (PC) model. While effective for perceiving near-threshold stimuli, these strategies can lead to the integration of irrelevant information. This study elicited the sound-induced flash illusion (SiFi) in younger and older adults while recording magnetoencephalography (MEG). Older adults perceived more illusions and exhibited increased pre-stimulus beta-band activity, a neural marker of top-down predictions. Connectivity analysis revealed stronger illusion-related modulation from auditory to visual cortices in older adults, suggesting an increased reliance on cross-modal predictive templates.
The Challenge of Sensory Aging
As we age, our primary senses—vision and hearing—naturally degrade. To maintain a coherent perception of the world, the brain must compensate. One theory, Predictive Coding (PC), suggests the brain acts as a "Bayesian inference machine," balancing incoming sensory evidence with "priors"—internal models built from a lifetime of experience. This study investigates whether older adults rely more heavily on these internal priors than younger adults, a strategy that helps navigate weak stimuli but may increase susceptibility to multisensory illusions.
Methods: Testing the Sound-Induced Flash Illusion (SiFi)
The study compared 25 younger adults (average age 27) and 28 older adults (average age 67). Participants performed the Sound-Induced Flash Illusion (SiFi) task. In this paradigm, a single flash of light is presented alongside two quick beeps. The brain often "fills in" a second flash, leading to an illusory percept.
Researchers utilized Magnetoencephalography (MEG) to record brain activity with millisecond precision. To analyze information flow between brain regions, they employed Transfer Entropy (TE) and Dynamic Causal Modeling (DCM), testing specific physiological connectivity models against the observed data.
Key Insights: The Role of Beta Oscillations
In predictive coding theory, beta-band oscillations (12–30 Hz) are thought to carry "top-down" predictions from higher-order brain areas to primary sensory regions. The study found that older adults exhibited significantly higher beta-band activity before the stimuli appeared. This suggests their brains were "pre-loaded" with stronger expectations compared to younger adults.
Furthermore, the illusion appears to stem from cross-modal compensation. In older adults, the connectivity from the auditory cortex to the visual cortex was more heavily modulated during the illusion. Essentially, the auditory system "instructs" the visual system on what it should be seeing based on the auditory input.
Key Results: Behavioral and Neural Findings
- Behavioral Performance: Older adults perceived significantly more illusions than younger adults, particularly as the Stimulus-Onset Asynchrony (SOA) between beeps increased (100–500 ms).
- Neural Activity: Older adults showed higher pre-stimulus beta-band power (from -250 to 75 ms relative to stimulus onset).
- Connectivity: DCM analysis revealed that while the basic brain connectivity model was similar across groups, the strength of modulation differed. Older adults showed significantly stronger links from the primary auditory cortex to the fusiform gyrus (a visual processing area) during illusory trials.
Limitations and Future Directions
The researchers noted several considerations for future work:
- Sample Heterogeneity: Older adults vary significantly in sensory acuity and cognitive strategy, which introduces variability in the data.
- Fatigue Management: To prevent exhaustion, fewer trials were conducted with older participants, potentially impacting the statistical power of the MEG signal.
- Causal Links: While the study demonstrates a strong correlation between pre-stimulus beta activity, connectivity, and the illusion, it does not prove causality. Future research using non-invasive brain stimulation (e.g., TMS) could manipulate these oscillations to determine if they directly control illusion rates.
Summary updated