Theta Oscillations as a Bridge between Perception and Memory

In natural speech comprehension, listeners rely on more than just the sound of a speaker's voice. Visual cues, such as facial expressions and lip movements, significantly enhance understanding, especially in noisy environments or when speech is fast and complex (Peelle and Sommers, 2015). As many academics may notice, watching the speaker during a lecture, instead of focusing solely on the slides, can improve both attention and comprehension. These observations highlight the importance of visual information in supporting auditory processing and the brain's ability to integrate multisensory inputs during language comprehension. But how does the brain optimally bind visual and auditory information into a coherent speech perception? One proposed mechanism is that aligned sensory inputs, reflected in synchronous cortical oscillations, are more effectively associated in the hippocampus through theta-phase-modulated plasticity.

Speech has a natural temporal structure in which syllables typically occur at a rate of ∼4–8 Hz. Because lip movements align closely with syllable onsets, both visual and auditory speech signals are naturally organized at this timescale. As a result, neural activity in auditory and visual cortices tends to follow the stimulus rhythm, reflecting entrainment to the temporal structure of speech (Luo et al., 2010; Park et al., 2016). Notably, this frequency range overlaps with the theta band, which has been widely implicated in mnemonic processes in the hippocampus. Across species, hippocampal theta oscillations are associated with episodic memory encoding, attention, and navigation, and their phase modulates synaptic plasticity mechanisms that support memory formation (Jacobs, 2014). Synchronization between cortical inputs and hippocampal theta is therefore proposed to offer a temporal framework that favors both short-term information maintenance and longer-lasting plastic changes, providing a plausible mechanism by which multisensory speech signals may be integrated into memory.

One reason theta rhythms may be especially effective in organizing perception and memory is their modulation of synaptic plasticity. Long-term potentiation (LTP), a cellular mechanism thought to underlie memory formation, is enhanced when inputs arrive at specific phases of the theta cycle (Hyman et al., 2003). Conversely, inputs at other phases can induce long-term depression. Thus, theta provides a temporal structure that selectively amplifies coincident signals while suppressing others, reinforcing the notion that oscillatory synchrony across brain regions may support memory by coordinating the timing of neural activity (Fell and Axmacher, 2011).

In a recent magnetoencephalography study, Biau et al. (2025) investigated whether theta-phase alignment between visual and auditory speech signals improves memory encoding and retrieval. Participants watched 5 s video clips of a speaker articulating short sentences and tried to classify them into two conditions. While the synchronous condition maintained the natural alignment of inputs, the asynchronous condition featured an auditory stream delayed by 180° (antiphase) relative to the dominant theta component shared by lip movements and the sound envelope. Although participants demonstrated a response bias toward perceiving the clips as synchronous, the manipulation was intended to disrupt theta-phase coherence between auditory and visual inputs.

The authors formulated three main predictions. First, synchronous audiovisual speech should be associated with increased theta power in auditory, visual, and hippocampal regions during encoding. Second, theta-phase patterns established during encoding should be reinstated during retrieval. Third, if theta-phase alignment supports memory encoding, behavioral memory performance should be superior in the synchronous condition.

As expected, synchronous encoding was associated with higher theta power in neocortical and hippocampal regions compared with the asynchronous condition. Moreover, theta-phase patterns observed during encoding were reinstated during retrieval, particularly following synchronous stimulation. Nevertheless, memory performance for speech content did not differ significantly between conditions. Together, these results suggest that theta-phase alignment during perception influences the neural stability of memory traces, even when behavioral effects are not detectable.

How should these results be interpreted? The lack of a behavioral effect does not necessarily undermine the neural findings. The speech clips were brief and contained meaningful content, which may have limited the task's sensitivity to detect differences in memory performance. While memory consolidation relies on long-term mechanisms, such as LTP, speech comprehension also engages short-term retention processes. Understanding a sentence requires maintaining earlier words in memory until the sentence's conclusion. This form of short-term processing—often referred to as short-term plasticity (STP)—could theoretically benefit from theta-phase synchronization, although this was not directly tested in the present study. Synchronized audiovisual input could support the transient maintenance of syllabic and lexical information over short timescales, thereby facilitating comprehension during unfolding speech. To determine whether these mechanisms can produce measurable differences in recall, future studies should employ more conceptually demanding and temporally extended speech materials.

Rodent research supports this dual function of theta in both memory formation and online information processing. In rodents, theta rhythms coordinate activity between the hippocampus and neocortex during active behaviors, enabling the integration of information across sensory domains. Theta also modulates interneuron firing and pyramidal output in a phase-specific manner (Klausberger and Somogyi, 2008). These dynamics support both the encoding of novel experiences and the coordination of ongoing sensory input—functions that may have analogs in the human brain during speech processing.

Cross-species comparisons, however, require caution. In humans, theta activity linked to associative memory formation is typically observed in a lower frequency range 2–5 Hz. In contrast, rodents exhibit faster hippocampal theta oscillations in the 4–10 Hz range, which are functionally segregated by phase to support encoding and retrieval processes. These frequency differences may reflect species-specific variations in anatomical scale and conduction delays (Jacobs, 2014). While rodent hippocampal theta is known to depend on circuits involving septal input and layer-specific activity, the cellular sources and circuit-level origins of human theta remain unresolved, raising important questions about whether the same mechanisms are at play. Defining oscillations solely by frequency may overlook important differences in their cellular and network-level bases across species (Lopes-dos-Santos et al., 2025).

Although we must avoid assuming that human theta is analogous to rodent theta, the functional parallels are compelling. Both appear to support multisensory integration, temporal binding, and memory formation. Biau et al.'s findings extend this framework by showing that synchrony between visual and auditory speech cues modulates theta activity in humans, particularly in regions involved in memory encoding and retrieval. Future work could test whether larger asynchronies or more complex, naturalistic speech materials, including those cadenced out of theta range, elicit clearer behavioral differences. Tasks that tax working memory over longer durations may be more sensitive to disruptions in audiovisual timing and may help disentangle the contributions of STP and LTP.

In summary, the study by Biau et al. reinforces the idea that theta oscillations act as a temporal bridge between perception and memory. By coordinating activity across cortical and hippocampal regions, theta rhythms provide a shared timing framework that organizes neural activity across sensory and associative networks. This coordination may underlie both speech comprehension and memory formation, especially when sensory inputs are temporally aligned—as occurs in natural speech, where the dynamics of lip aperture and sound envelope coincide for the listener. While differences between human and rodent theta must be acknowledged, the cross-species convergence on this mechanism points to a potentially conserved role in neural computation, offering valuable insight into how the brain weaves experience into coherent thought.