Transformer-based deep learning models have achieved state-of-the-art performance across numerous language and vision tasks. While the self-attention mechanism, a core component of transformers, has proven capable of handling complex data patterns, it has been observed that the representational capacity of the attention matrix degrades significantly across transformer layers, thereby hurting its overall performance. In this work, we leverage the connection between selfattention computations and low-pass non-local means (NLM) smoothing filters and propose the Twicing Attention, a novel attention mechanism that uses kernel twicing procedure in nonparametric regression to alleviate the low-pass behavior of associated NLM smoothing with compelling theoretical guarantees and enhanced adversarial robustness. This approach enables the extraction and reuse of meaningful information retained in the residuals following the imperfect smoothing operation at each layer. Our proposed method offers two key advantages over standard self-attention: 1) a provably slower decay of representational capacity and 2) improved robustness and accuracy across various data modalities and tasks. We empirically demonstrate the performance gains of our model over baseline transformers on multiple tasks and benchmarks, including image classification and language modeling, on both clean and corrupted data. They have also demonstrated strong performance in knowledge transfer from pretraining tasks to various downstream tasks with weak or no supervision (Radford et al., 2018; 2019; Devlin et al., 2018). At the core of these models is the dot-product self-attention mechanism, which learns self-alignment between tokens in an input sequence by estimating the relative importance of each token with respect to all others. The mechanism then transforms each token into a weighted average of the feature representations of the other tokens with weights proportional to the learned importance scores.