Professor Miguel Nicolelis (2019) has published a free copy of his contributions to BMI (brain-machine interfaces) emphasizing his twenty years of work starting in 1999 and continuing through 2015.* Until 2003, Nicolelis had no competitors, but shortly thereafter Andersen et al. (2003), Schwartz et al. (2004) and Donoghue et al. (2006) joined the field, and tried to eclipse him and his associates [as described in Tehovnik, Waking up in Macaíba, 2017]; they, however, failed to achieve the eclipse, since the information transfer rate of their devices were typically below 1 bit per second at an average of about 0.2 bits/sec, much like what Nicolelis’ devices were transferring (Tehovnik and Chen 2015; Tehovnik et al. 2013). By comparison, the cochlear implant transfers 10 bits/sec (Tehovnik and Chen 2015) and therefore has been commercialized with over 700,000 registered implant recipients worldwide (NIH Statistics 2019).
BMI technology is still largely experimental. Willett, Shenoy et al. (2021) have developed a BMI for patients that transfers up to 5 bits/sec for spontaneously generated writing, but it is unclear whether this high rate is due to the residual movements (Tehovnik et al. 2013) of the hand contralateral to the BMI implant. To date, the most ambitious BMI utilizes a digital bridge between neocortex and the spinal cord below a partial transection to evoke a stepping response that still requires support of the body with crutches; but significantly the BMI portion of the implant in M1 enhances the information transfer rate by a mere 0.5 bits per second, since most of the walking (86% or 3.0 bits/sec of it) is induced by spinal cord stimulation in the absence of the cortical implant (Lorach et al. 2023). Accordingly, BMI falls short of the cochlear implant and thus BMI developers are years away from a marketable device. The pre-mature marketing by Nicolelis at the 2014 FEFA World Cup of his BMI technology (Tehovnik 2017b) should be a warning to Elon Musk (of Neuralink) that biology is not engineering, for if it were a BMI chip would now be in every brain on the planet. See figure that summarizes the information transfer rates for various devices including human language.
The amount of information current-day BMI systems transfer varies depending on the type of BMI and the specific task being performed. Here's a breakdown:
Information transfer rate (bits/second):
- EEG-based BMIs (non-invasive): These BMIs measure electrical activity from the scalp and generally have lower information transfer rates, ranging from 0.25 to 0.5 bits/second. This is enough for basic control tasks like cursor movement or simple word selection, but not for complex actions.
- Invasive BMIs: These BMIs use electrodes implanted directly in the brain, providing access to more detailed neural signals. Information transfer rates can be higher, reaching up to 40 bits/second for simple tasks like motor control. However, this is still significantly slower than natural human communication rates, which can reach 40-100 bits/second for speech and even higher for complex forms of communication like writing.
Factors affecting information transfer rate:
- Type of brain activity: Different brain areas and signals carry different amounts of information. Motor cortex activity used for cursor control is easier to decode than complex cognitive processes like thoughts or emotions.
- Electrode technology: The number and placement of electrodes influence how much neural activity is captured. More electrodes and better placement can lead to higher information transfer rates.
- Signal processing algorithms: Algorithms used to interpret and decode brain signals play a crucial role in extracting information. Advancements in machine learning and artificial intelligence are improving decoding accuracy and information transfer rates.
Current limitations:
- Low information transfer rates: Compared to natural communication, current BMIs are still relatively slow and limited in the complexity of information they can transfer.
- Accuracy and reliability: Decoding brain signals can be challenging, leading to errors and inconsistencies in control.
- Ethical considerations: Invasive BMIs raise ethical concerns about privacy, security, and potential misuse of brain data.
Despite these challenges, BMI research is rapidly advancing, and information transfer rates are expected to improve significantly in the future. This could revolutionize various fields, including healthcare, rehabilitation, and human-computer interaction.
Dear Dr. Murtadha Shukur:
The following statement is inconsistent with the literature on penetrating recording micro-electrodes for primates including humans:
"These BMIs use electrodes implanted directly in the brain, providing access to more detailed neural signals. Information transfer rates can be higher, reaching up to 40 bits/second for simple tasks like motor control. However, this is still significantly slower than natural human communication rates, which can reach 40-100 bits/second for speech and even higher for complex forms of communication like writing." (Shukur Dec 14, 2023, ResearchGate)
And the aforesaid should only be accepted once the peer-reviewed data and calculations to yield the 40 bit per sec are provided.
Note, many of the values in the posted figure were calculated using Shannon's equations, and these values have passed peer view (for details see: Tehovnik and Chen 2015; Tehovnik et al. 2013; Tehovnik and Teixeira e Silva 2014).
Also, for current estimates of information transfers for human language including English, Basque, Turkic, Vietnamese, Japanese, etc. , see:
Coupé, C., Oh, Y.M., Dediu, D., Pellegrino, F., 2019. Different languages, similar encoding efficiency: Comparable information rates across the human communicative niche. Science Advances 5, eaaw2594. (Paper Attached).
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Materials Science
- Plating