In this paper, we present a novel communication channel, called the absorption channel, inspired by information transmission in neurons. Our motivation comes from in-vivo nano-machines, emerging medical applications, and brain-machine interfaces that communicate over the nervous system. Another motivation comes from viewing our model as a specific deletion channel, which may provide a new perspective and ideas to study the general deletion channel. For any given finite alphabet, we give codes that can correct absorption errors. For the binary alphabet, the problem is relatively trivial and we can apply binary (multiple-) deletion correcting codes. For single-absorption error, we prove that the Varshamov-Tenengolts codes can provide a near-optimal code in our setting. When the alphabet size $q$ is at least $3$, we first construct a single-absorption correcting code whose redundancy is at most $3\log_q(n)+O(1)$. Then, based on this code and ideas introduced in \cite{Gabrys2022IT}, we give a second construction of single-absorption correcting codes with redundancy $\log_q(n)+12\log_q\log_q(n)+O(1)$, which is optimal up to an $O\left(\log_q\log_q(n)\right)$. Finally, we apply the syndrome compression technique with pre-coding to obtain a subcode of the single-absorption correcting code. This subcode can combat multiple-absorption errors and has low redundancy. For each setup, efficient encoders and decoders are provided.