Findings Log is a look at some of the research and academic papers that have recently caught the eyes of Seed’s editors, Lee Billings, Joe Kloc, and Maywa Montenegro. For more recommended reading and occasional insights, follow them on Twitter.
Read the paperJ.J. Carabias-Orti et al. IEEE Transactions on Audio, Speech, and Language Processing. March, 2010
Music for the Masses
The American folklorist Alan Lomax spent much of the 20th century traveling around the world, and most notably the American South, recording thousands of folk and blues songs. These songs—now kept in the Library of Congress—are traces of a time when recording technology was just beginning to develop. For this reason, learning to play the songs is difficult: Many have never been accurately transcribed (or at least no accurate transcription has been published), and the music in the crackly, century-old recordings is often difficult to decipher. To anyone who has spent any amount of time attempting to play these songs, the same thought always occurs: Shouldn’t there be a way to parse them automatically?
The short answer is that there is—or rather, there will be soon. Computer scientists have been making great strides in the past few years in developing tools that can take an acoustic signal—a song—and “decompose” it into musical notes. In essence, what these programs do is match elements of an audio signal to entries in what is known as a “harmonic dictionary.” This “translation”—or perhaps more accurately, decomposition—allows the song to be converted into sheet music. But a major problem with many of these decomposition methods has been that as factors like musical style, performer, and instrument change, a new dictionary is required. Now, in a recent article in IEEE Transactions on Audio, Speech, and Language Processing, researchers have developed a dictionary that can automatically “adapt” to these different factors. This means, in theory, that the same tool used to transcribe a jazz composition could be used to transcribe an old blues record. Currently the method is limited to musical recordings containing only one instrument, but the work is an important step towards devising an accurate way to automatically convert audio recordings into accurate sheet music. Perhaps someday soon the vast stores of Lomax’s recordings will be truly accessible to amateur musicians everywhere.—JK
Read the paperNational Academy of Sciences.
The Dose Makes the Poison
The verdict is in. After decades of piecemeal studies on genetically engineered crops—some showing promise, others showing pitfalls—the US National Research Council has published a first-ever comprehensive assessment of the environmental and economic impacts of GE crops on American farmers. In a 200-page report issued earlier this month, the panel found that, when compared to conventional crops, GE varieties have provided “substantial” gains. They also included some caveats, however, that may be damning enough to overshadow the good news.
Sifting through reams of peer-review studies, the NRC identified several clear-cut biotech benefits: Corn enhanced with a bacterial gene for an insect-killing toxin, for example, enables farmers to cut down on pesticide usage, saving them the expense of chemicals and protecting other farm-dwelling wildlife. Soybeans re-jiggered with a gene that gives them immunity to the herbicide glyphosate reduce the need to till fields, resulting in less soil erosion—which in turn means that less sediment, fertilizer, and pesticides wash into streams. And although overall herbicide use has increased due to mass planting of glyphosate-resistant crops, the glyphosate chemical—better known by its trademarked name “Roundup”—has generally replaced more toxic herbicides.
On a less sanguine note, however, the NRC also found that the benefits of GE may be fleeting. Overuse of the glyphosate-resistant “Roundup Ready” crops is already beginning to result in glyphosate-tolerant weeds, causing farmers to use additional herbicides, many of which are more toxic than glyphosate. Also, the prices of biotech seeds have risen sharply in the past few years. The study examined data from the early years of GE use, so the economic benefits to farmers it identified may not fully reflect current realities.
Given these ambiguities, biotech proponents and critics alike are likely to find support for their views in this report. One important thing to keep in mind, however, is that the NRC only compared genetically modified crops to conventionally grown ones, not to organic crops. Some scientists, like University of California-Davis geneticist Pamela Ronald, suggest that the biggest gains, both for farmers and the environment, would come from a merger of organic farming and GE seeds. With early signs indicating that biotech crops will undermine their own effectiveness when sown across vast monoculture landscapes, let’s hope it doesn’t take decades for scientists to analyze this provocative idea.—MM
Read the paperOwen et al. Nature, advance online publication.
April 20, 2010
Each year, consumers spend millions of dollars (and untold hours) on “brain-training” software, products offering computerized tests and drills that supposedly sharpen and strengthen one’s thinking through frequent use. The brain, so the pitch goes, is akin to musculature in its response to repetitive, simple activity: Just as leg presses can improve one’s vertical leap and running speed while also toning and swelling one’s quadriceps, surely the brain can glean cognitive benefits from tasks like word puzzles and memorizing strings of digits. The only problem is that this assumption hasn’t been thoroughly tested; whether brain-training activities generally boost cognitive skills in addition to improving specific task performance has been unclear.
Now, a research collaboration between British scientists and the BBC Lab UK website offers evidence that brain training isn’t as effective as advertised. The researchers enlisted viewers of the popular science program, Bang Goes the Theory, to take part in a six-week study examining the effects of brain training. More than 11,000 participants between the ages of 18 and 60 completed the online study, which split volunteers into three different groups after administering benchmark aptitude tests. One group was presented with general tasks related to reasoning, problem solving, and planning; the second performed typical brain-training tasks that required short-term memory, math, attention, and visuospatial processing. A third group simply consulted the Internet to answer a series of obscure questions. The difficulty of all tasks for all groups was increased over time as participants became more adept, thus providing a continuous challenge. At the end of the six-week period, the volunteers took benchmark tests again, and their previous scores were compared. The results: Volunteers in all three groups saw performance increases, but the brain-training group did not exhibit significantly greater increases to general cognitive function than the other groups. For the brain, practice makes perfect, but there may be no particular mental task ideally suited for sharpening general cognition.
The study is far from definitive, for several reasons. Brain training is often pitched as most beneficial for the very young, the very old, and individuals recuperating from debilitating brain injuries. A self-selected group of healthy young adults and mid-lifers could perhaps be less responsive. Further, some critics of the study say it didn’t allow enough time for the positive effects of brain-training activities to fully manifest. The work will undoubtedly spur more studies. In the meantime, it’s probably best to take brain-training claims with a grain of salt, and to remember the clear evidence that old-fashioned vigorous physical activity provides distinct cognitive benefits.—LB
Originally published April 27, 2010